Standardization and Coding of Gastrointestinal Endoscopy ... Marcel Johannes Maria.pdf · Standardization and Coding of Gastrointestinal Endoscopy Reports Standaardisatie en codering

Standardization and Coding of Gastrointestinal Endoscopy reports

Standardization and Coding of

Gastrointestinal Endoscopy Reports

Standaardisatie en codering van gastro-intestinale endoscopie verslagen

Proefschrif t

ter verkrijging van de graad van doctor aan de Erasmus Universiteit Rotterdam op gezag van rector magnificus

Prof.dr. H.G. Schmidt

en volgens besluit van het College voor Promoties.

De openbare verdediging zal plaatsvinden op vrijdag 15 april 2011 om 13.30 uur door

Marcel Johannes Maria Groenen

geboren te Tilburg

Promotiecommissie

PromotorProf.dr. E.J. Kuipers

Overige ledenProf.drs. J.F.W.M. BartelsmanProf.dr. C.H.J. van EijckProf.dr. H.W. Tilanus

CopromotorDr. R.J.Th. Ouwendijk

5

Contents

Chapter 1 General introduction and outline of the thesis 7

Chapter 2 Computerization of endoscopic reports using standard reports and text-blocks 27

Chapter 3 Computerized endoscopic reporting is time competitive with conventional methods 43

Chapter 4 A cost–benefit analysis of endoscopy reporting methods: handwritten, dictated and computerized 55

Chapter 5 Gastrointestinal Endoscopic Terminology Coding (GET-C): a WHO-approved extension of the ICD-10 71

Chapter 6 Validation study of automatically generated codes in colonoscopy using the endoscopic report System Endobase® 83

Chapter 7 Incidence of duodenal ulcers and gastric ulcers in a western population; back to where it started 95

Chapter 8 Trends in the diagnostic yield of open access upper gastrointestinal endoscopy 109

Chapter 9 125 SummaryandDiscussion 127 Samenvattingendiscussie 133

apppendix 139 Dankwoord141 CurriculumVitae145

7

General introduction

and outline of the thesis

Chap ter 1

9

Gastrointestinal endoscopy

Gastrointestinal endoscopy has developed rapidly in the last five decades. It started in the sixties, when the first commercial model of the then recently invented fully flexible fiberoptic gastroscope was developed.1 In the late sixties fiberoptic endo-scopes changed to forward viewing, with an open channel for air insufflation, aspiration and a passage for accessories, especially biopsy instruments.2 In the seventies with the introduction of a new side-viewing endoscope it became possible to visualize the pancreatic duct and to perform endoscopic sphincterotomy, later accompanied by stone extraction.3 Enteroscopy was first established in the mid seventies and advanced into balloon-assisted enteroscopy, which enabled visual-izing and treating the whole small intestine in the new millennium. Endoscopic ultrasound (EUS), developed in the 1980’s, has in recent years become tool for diag-nosis and therapy of a range of esophageal, gastric, hepatobiliary, pancreatic and rectal disorders.4

Nowadays these innovative techniques have evolved into a routine investiga-tion of the gastrointestinal tract. Patients with gastrointestinal complaints gen-erally undergo an endoscopic examination. In the Netherlands yearly approxi-mately 400,000 examinations of the gastrointestinal tract are performed. With these endoscopic studies abnormalities in esophagus, stomach, duodenum, colon, small intestines and bile ducts are diagnosed and therapeutic interventions are performed. The findings of these investigations have important implications for patient management, for example for the selection of medical therapy for a gas-tric ulcer or for the indication for a surgical intervention of a malignant process. The endoscopist generates a report of the performed examination for the referring physician. The reports range from short written or dictated reports to standardized computer reports.

As endoscopy is primarily a real-time visualizing investigation, it is of extreme importance to report the diagnostic findings and therapeutic interventions in well understood terminology to others. Guidelines from different gastrointestinal endoscopy societies have for that purpose been published to specify minimal stan-dards for an endoscopy report.5-6 These however only describe the elements that are required in a report, not the way how a report has to be composed.

The Quality Committee of the Dutch Society of Gastroenterologists and Hepa-tologists (NvMDL) published a report in 2005 entitled ‘Tijd voor Kwaliteit’ (Time for Quality). In this report quality standards of the Society were defined. These stan-dards can be seen as a part of a quality system for gastroenterologists, designed to establish, test and improve quality of care. It was stated that the results of an endoscopic investigation should be reported in written and/or digital format. Com-pulsory elements of the report are listed in Table 1.

1

10

Chap ter 1

Traditionally most of the reports were handwritten, but since 1980 several endo-scopic information systems have been developed to record endoscopic findings, to store images and to assist in report composition.

Quality indicators for gastrointestinal endoscopic procedures

Quality assurance in endoscopy has taken increased importance. This was further enhanced by the introduction of colorectal screening programs started in various countries and the demonstration that the quality of bowel preparation and colono-scopy had a major influence on the prevention of cancer.

High quality endoscopy requires that the patient receives a well indicated pro-cedure, that correct and clinically relevant diagnoses are made, that therapies are properly performed, and that all is accomplished with minimum risk for the pa-tient. To provide patients with the best possible care, quality indicators for endo-

Table 1 Eelementsthatshouldbementionedinareport

• Date,eventuallytimeofexamination• Typeofendosopicexamination• Patientcharacteristics(Name,birthdate,registrationnumbers)• Referringdoctor• Indication• Setting• ClinicalversusOutpatient/ambulatoryprocedure• Electiveversusemergencyprocedure• Location(Endoscopyunit,intensivecareunit,operatingroom)• Relevantmedication• Informedconsent• Pre-medication• Monitoring(oximetry,bloodpressure,pulserate)• Administeringoxygenormedication(e.g.flumazenil,atropine)• Equipmentused(typeofendoscope)• Proceduralaspects(discomfortforpatient,processofexamination)• Findingsanddiagnosticprocedures• Therapeuticinterventionsandmaterialused• Biopsies,punctures,samples• Imagecaptureandimagestorage• Complications• Advicetoreferringdoctorfollowingtheexamination• Information,advice,prescriptionsgiventothepatient• Nameofendoscopist

QualityReport‘TijdvoorKwaliteit’DutchSocietyofGastroenterologistsandHepatologists(NvMDL)

11

Gener al introduction and outline

scopy have been developed. These indicators may be used to improve overall quality of endoscopy services.

Two quality indicators for endoscopy units are specified in the ‘Basisset zieken-huizen kwaliteitsindicatoren 2011’ from the Dutch Healtcare Inspectorate (IGZ).7 The first indicator refers to the 7 × 24 hour availability of a trained team to per-form intervention endoscopy. The second indicator comprises of the execution of an emergency endoscopy within 24 hours in case of bleeding from the upper gastroin-testinal tract.

Besides these accepted quality indicators, a number of further indicators have been proposed by endoscopy societies.8-9 These can be divided in pre-procedure , pro-cedure, and post-procedure indicators. Pre-procedure indicators address the indica-tion, medical history, risk assessment, and patient consent. These indicators must be documented before any procedure has begun. There must be a proper indication that the information gained or the therapy provided will help the patient. The patient must be properly informed about the most common complications, and the patient’s consent must be obtained and filed. The patient’s medical history must be recorded and it is important to notice prior gastrointestinal surgery, implanted defibrillators, previous adverse experience with sedation, drug allergies etc. The use of anticoagulants or antiplatelet medications has to be recorded. Risk strati-fication should be performed before sedation is begun. Prophylactic antibiotics should be administered to high risk patients in high risk procedures. The intended level of sedation should be specified before the procedure. A time-out procedure to confirm the correct patient and the correct procedure should be performed

Procedural indicators include photo-documentation, in which images of major abnormalities and landmarks like coecum are recorded; patient monitoring dur-ing sedated endoscopic procedures comprising pulse oximetry, pulse rate and blood pressure; recording of administered doses and route of the medication, including any reversal agents like flumazenil or naloxon when applied. Post-procedure indi-cators include documentation of the discharge from the endoscopy unit. The pa-tient should be provided at discharge with written instructions and with informa-tion regarding the nature of potential delayed complications and how to act upon them. A procedure report should be prepared immediately after the procedure, con-taining the elements of the endoscopic procedure that are listed in the different guidelines.10 Complications like adverse events or unplanned interventions should be reported by the endoscopy unit. Patient satisfaction should be collected by a vali-dated questionnaire. The eventually resumption of anticoagulants or antiplatelet medication should be recorded. The complete report comprising the endoscopic findings, therapy and follow-up recommendations should be stored for immediate retrieval if needed, and sent to the referring doctor.

Most mentioned quality indicators are directly related to the endoscopy report.

12

These can be conveniently recorded using advanced computerized endoscopy report systems.

Computerization of endoscopy reports

During the last decades various computerized systems have come available for medical reporting including for gastrointestinal endoscopy.11-23 Most of these sys-tems were stand-alone report-systems, designed to compose a report of the per-formed examination. Some systems have also the possibility to store images and videos taken during the endoscopy. It is essential to store the endoscopy images DICOM (Digital Imaging and Communications in Medicine) compatible in order to enable their export to other sources. Recording digital images is of vast importance for medical quality control, therapy evaluation, and education purposes.

Computerization of endoscopy reports has many important aspects. When only the report is generated, it is necessary to link the system to the Hospital Informa-tion System (HIS), making the report available for every authorized doctor at any time. These links exists in two directions. One link runs from the HIS to the re-port system transporting patient characteristics like name, gender, birth date, ad-dress, insurance number, name of general practitioner, risk factors. In the other direction the report is sent form the report system to the HIS and eventually directly to the referring doctor via electronic means. In this way the report can be assessed 24 hours a day and in a good readable format.

Endoscopic electronic medical record systems to serve endoscopy units

Besides their central role of generating the endoscopy procedure report, endoscopic electronic medical record systems (EEMR’s) have also evolved into sophisticated da-tabases with extensive possibilities which all add to practice management, patient care, risk reduction, research purposes and so lead to continuous quality improve-ment of the endoscopic practice.

With connection to the HIS, any risk factors and allergies of patient can be eas-ily recalled. Further functions may include interfaces with other medical systems such as pathology databases and direct electronic communication with the refer-ring physicians.

The system can be used for patient scheduling and automated follow-up. More recent features are the possibility to track inventory, such as drugs, endoscope us-age and endoscopic accessories. The system allows generation of endoscopy unit

Chap ter 1

13

productivity statistics, like involvement of nursing and technician personnel. Physicians and trainees can also monitor their procedure volumes and proposed quality indicators like coecum intubation and polyp detection rates. These features can assist to streamline practice management.

In these systems it is also possible to store nursing information before, during and after the investigation, which can be included the patient care document.

Complications that appear during the endoscopic examination, direct after or some days later has to be recorded in the system. Also when sedation is given pa-tient vital sign monitoring, with oxygen saturation, heart rate and blood pressure, has to be recorded according to guidelines. Intraprocedure medication can be en-tered in the system and transferred automatically to the report.

The information of the endoscope used in a patient has to be stored. In some systems the endoscope is automatically recognized by a chip in the endoscope at the moment the endoscope is connected to the processor. At the moment that the patient data are connected to the same processor, the endoscope is coupled to the patient in the EEMR. Information of the cleaning process and subsequently drying process is stored in the same way in the EEMR. In this way a complete track and tracing of the endoscope is guaranteed and the risk for infection is hereby mini-mized. In case of a not correctly cleaned or stored endoscope, a message is given at the time of selection a new patient to act upon prior to start of the procedure.

During composition of the endoscopy report diagnostic codes can be automati-cally generated. These codes are based on indications, findings and therapeutic in-terventions that are selected. Furthermore, procedure codes for financial registra-tion can also be automatically produced.

Because the data of former investigations are stored in the EEMR, it is easy to re-view reports, images and eventually videos of endoscopic investigations performed in the past. Pre-medication given at earlier investigations can prophesy the needed dosage of pre-medication for a new endoscopy.

Coding of data

The data gathered in the electronic medical record is not always appropriate for re-search purposes. Because different ways of report writing are available it is hard to compare these different data. Besides this it is nearly impossible to compare the endoscopic data produced by diverse systems because of dissimilar semantics.

To enhance the consistency and comparability of endoscopic reports we intend to develop a comprehensive WHO-approved code system for gastrointestinal en-doscopic terminology. The International Classification of Diseases, 10th edition (ICD-10), and the ICD-10 clinical modification (ICD-10-CM) were used as basis and


14

expanded to allow description of every possible gastrointestinal endoscopic term under conditions defined by the WHO.

The new extended code system was named Gastrointestinal Endoscopic Termi-nology Coding (GET-C). The GET-C is incorporated in the electronic report system En-dobase that is used in the trans.it project so that endoscopic data, produced in different ways of report writing, are coded and consequently comparable.

Trans.it Project

In 1998, ten Dutch hospitals started a collaboration entitled trans-it, seeking to further develop and improve the uniform standardized reporting of gastrointes-tinal endoscopic findings. All Trans.it hospitals use the same report system, En-doalpha Documentation (former Endobase), with standard reports, text blocks, and structured input. Three academic hospitals (Amsterdam Medical Center, Eras-mus Medical Center, University Medical Center Utrecht), and 7 district hospitals (Atrium Hospital Heerlen, Elisabeth Hospital Tilburg, Ikazia Hospital Rotterdam, Meander Medical Center Amersfoort, Rijnstate Hospital Arnhem, Tergooi Hospital Blaricum, Vlietland Hospital Vlaardingen) participate in the Trans.it work group.

From each of these hospitals at least one gastroenterologist has a permanent seat in the Trans.it project group. The work group meets regularly to discuss new developments and improvements. Within the work group, the members work con-tinuously on improving the quality of the standard reports and text blocks. It is partly in response to comments and recommendations from the Trans.it project group that the Endobase program has been continuously modified to accommodate the requirements of those using it. All standard reports and text blocks have been discussed and accepted by the members of the work group.

Users of Endobase can also recommend changes to the standard reports and text blocks. No modification can be introduced without the agreement of the majority of the Trans.it group. Immediately after acceptance, the modifications are made by remote control in all Trans.it hospitals. This guarantees the necessary unifor-mity in report writing.

Central database of Trans.it

One goal of the Trans.it Project is to build a central database with anonymous data of endoscopy investigations from the participating hospitals, which jointly carry out some 30,000 to 40,000 examinations annually. This gives rise to a large data-base with an abundance of information on endoscopic examinations. Thanks to

Chap ter 1

15

the uniform report system and use of GET-C coding, it is possible to analyze the data anonymously. After a few years, it will be possible to obtain important information on the epidemiology of various disorders and the effects of diagnostic and thera-peutic intervention for doctors and policy makers in the healthcare sector.

In order to save the data in a central database, a safe communication structure is needed, which can also anonymize the data. An export tool was developed for this purpose. This involves the following activities: export, anonymization, and encryption (coding).

During the export procedure, the patient and source ID is modified so that the patient can no longer be identified on the basis of the data. A new code is add-ed, which guarantees the anonymity of the patient whilst enabling the central database to keep various details of the same patient together. Finally, to safeguard the confidentiality of the data, all information is encrypted before sending. This means that during the transfer of data the information can be coded in such a way that it cannot be read.

Description of Endobase/Endoalpha

The research in this thesis was performed on the endoscopic electronic medical record system Endobase. Endobase is a software solution for endoscopy units. It was developed in the 1980’s as a report module. In the beginning of the nineties the first version of Endobase was upgraded to version Endobase 2, in which simple standard reports and text blocks were added. In 2000 a complete new version with another


Figure 1 Procedureofexportinganonymousendoscopicdata

Hospital

E-mail

Centralserver

Endobase

AnonymizationEncryption

Register

PatientID Data+

hashedPatientID

Data+

hashedPatientID

Data+

16

database structure and extensive possibilities was launched, named Endobase III. The Trans.it work group started using this version and most of the research was done on this version of Endobase. The last years Endobase III developed in a program serving the endoscopy unit including an integration with other hospital systems which facilitate the practice management of the whole endoscopy unit. With this adaptation also the name changed to Endoalpha, the reporting module is named Endoalpha Documentation. In total 43 of the 92 Dutch hospitals have implemented Endoalpha Documentation for their gastroenterology departments. Besides gas-troenterology also pulmonology, surgery, gynecology, urology and ear, nose and throat specialists are using Endoalpha Documentation for reporting.

Scheduler and patient data and examination dataA scheduler module provides a planning for the whole endoscopy unit. The num-ber of available endoscopy rooms, the identification of the endoscopes, the required time for a specific examination and the number of available endoscopists can be put in the system listing endoscopy capacity per day. Preceding the examination the scheduler module plans each investigation taking these different variables in account.

Endobase uses a Health Level-7 protocol to communicate with other electronic medical systems to exchange medical data.24 At the moment that the patient per-sonal identification number is entered, all the available relevant personal, medical and administrative data can be extracted from the hospital information system. The data set contains personal data like name, address, birthday and gender, de-tails of the general practitioner; medical data about risks like allergies and previ-ous gastrointestinal surgery, including perform-date, relevant before starting the endoscopic examination; administrative date such as the name of the insurance company and the insurance number, useful for the financial administration, for which the data or procedure codes can be transmitted to the financial department by another HL-7 connection. Any alteration in the patient’s personal data or e.g. general practitioner will automatically be changed in the system.

After input of the patient data, the different examination data are selected. The user can define and choose between several kinds of examinations. The endosco-py-room where the examination is performed is selected. Next, you can select the referring doctor, the endoscopist and up to three residents and three endoscopy-assistants.

The indication of the examination is selected out of a definable table with one or more possible indications for that particular examination. These indications are linked to a specific GET-C code for research purposes. Medication given during the examination can be filled in beforehand from a defined list with recording of the dose given.

Chap ter 1

17

The kind of endoscope used and the internal number of this scope are automatically recognized and recorded in the system. At the moment the investigation is started, the start time and subsequently after finishing, the end time are being recorded. In the new version, certain landmarks, such as ileo-coecal valve, can be registered on time to be able to calculate the withdrawal time.

If the examination is an emergency, this option can be selected and it is stored the way the patient was referred for the endoscopy, by out-patient clinic or from within the hospital.

It is user-definable which fields are obligatory to fill in at the patient and exami-nation data input. This is an essential feature to be able to build a complete data-base appropriate for research purposes. Without given in all the obligatory data the user gets a message to fill in all the fields.

Examination and imagesDuring the examination, there is a direct connection between the Endobase com-puter system and the endoscope module in the endoscopy room. Patient charac-teristics are sent to the monitor in the endoscopy room. At the same time a video signal with the scene of the endoscope is transported to the system. The performed examination can be followed live from behind the computer.

During the endoscopy images can be captured by remote-buttons on the endo-scope. It is also possible to capture an image directly from behind the computer. From the endoscopy room the endoscopist can add voice notes to the images, so that these can be reproduced during the selection of the images. Besides still images also short movies can be stored by remote control. The resolution of the movies can be predefined.

All images and movies are stored as digital files in the system and will be avail-able for review and future use. After the examination up to twelve images (user definable) can be selected to print in the report. Failed images can be deleted to keep the database clean. In this field, it is also easy to compare images taken at earlier examinations. Results of therapy or alterations in severity of some chronic gastro-intestinal diseases can be easily visualized.

One can localize the position of the images in an anatomical picture of the gas-trointestinal tract. In this anatomical picture, also the localization and number of biopsies taken are recorded. This picture with the images is transported to the definitive report or can directly be printed together with the patient data.

For a better visualization of certain abnormalities in the images it is possible to edit the image in a draw module. One can enter some free text as comment, draw lines or circles and zoom in. Colors can be adapted and saved in a separate file, however the original image stays stored despite the changes made. Movies can be edited.


18

To build a personal database of noteworthy images you can copy some images in a user definable image-box. Because the images and movies are digitally stored, they are quickly exported and can be used in presentations or consultations of other specialists.

There is an option to create patient lists. If for example a patient is included in a trial, you can add the patient to such an own defined list.

After performing the examination it is possible to adapt and complete the ad-ministering of medication, changes of any endoscopes and diagnosis of former gastrointestinal operations of the patient.

When biopsies are taken, an order to the pathologist can be generated electroni-cally or in a paper report. Endoscopic images and the report can accompany the request for the pathologist.

Report writingThere are four different types of report writing in Endobase. Besides free text which is not usable for statistics, there are three other kinds of report writing from which can be chosen after performing the examination depending on the findings during the examination or the preference of the endoscopist.

First option is the standard report. Standard reports are based on diagnosis or com-bination of diagnosis. They contain the complete text describing the examination performed and all of the corresponding findings during this endoscopy.

This type of report is especially suited for about 30-40% of the endoscopic exami-nations in which no abnormalities are found.25

Chap ter 1

Figure 2 StandardreportsandTextblocks

StandardReportTextblockStomach

TextblockOesophagus

TextblockDuodenum

CompleteReport

19

Secondly, it is possible to compose a report by using text blocks. These reports are composed by selecting one or more diagnosis or findings from different sections of text-blocks. Each text-block consists of one or more sentences describing the se-lected diagnose or finding. The combination of these different diagnosis or sections concurrently creates the report. Almost every possible combination of findings can be composed with the use of these text-blocks.

Thirdly, Minimal Standard Terminology can be used for report writing. This mini-mal standard terminology is composed by a commission of the ESGE. Version 2.0 is used in the program.26 It was translated into Dutch and some extensions were made. Almost every part of the examination can be described by using this ter-minology. A sentence is build up out of different choices and all these sentences together compose the report.

All the three standardized ways of composing a report (Standard Reports, Text Blocks and Minimal Standard Terminology) are linked to GET-C. By selecting the diagnosis during the composing of the report, in the background, the correspond-ing codes are recorded in the database.


Figure 3 ScreenshotTextblocks

20

Besides diagnostic codes for research, procedure codes for financial aspects can be recorded. For a specific examination user-definable standards can be generated. Also manual selection of procedure codes out of a predefined list is possible. These codes are automatically transferred to the Hospital Information System and the financial department via HL-7 protocol.

Chap ter 1

Figure 5 Screenshotstructureddataentry

Figure 4 Structureddataentry

LumenMucosaAbnormalities



CompleteReportStomach

Esophagus

Duodenum

21

Patient history and external documentsIn the patient history all the endoscopic data of one specific patient are gathered. In this field one can quickly review the patients history of endoscopic examinations performed. The complete report of the examination is presented, and it is possible to view the captured images and movies of that particular examination. The char-acteristics of the patient and the particular examination are provided. In this way, the name of the endoscopist, the referring doctor and the used pre-medication of the former examination are directly accessible. All the patient characteristics like allergies and former operations are shown.

Any external document like histology results or X-ray images can be coupled to a specific examination. These results can be scanned by an attached photo-scanner or imported electronically from e.g. the hospital information system. In this way you can visualize the results of any taken biopsies.

This field is also suitable to register any possible complication that occurs. Com-plications occurring during the endoscopy need to be described in the report, but complications that appear later can be recorded at a specific examination in this history field.

With a browser based application it is allowed after authorization to view the complete patient file with all reports, images and videos throughout the hospital.

Patient care documentation and vital monitoring systemIn the Endoscopic Electronic Medical Record full patient care can be documented, like preparation protocol and informed consent form. The nurse can fill in any ex-amination documentation and discharge documentation.

Progressively sedation is given during endoscopy. Vital sign monitoring like, hart rate, blood pressure and oxygen saturation are mandatory. Registrations of these signs are automatically recorded in the system during the examination and after the examination in the recovery room.

StatisticsThe first Endobase 2 program was built on Interbase from Borland; Endoalpha uses Microsoft SQL database. These are convenient SQL Database programs. All data en-tered in the system can easily been extracted from the system for any purpose.

Endobase itself has a statistic module to execute extensive queries on different data-fields as well as on images. The results can be printed or exported. The ex-ported endoscopic data can be imported into a program especially made for statistic analysis like e.g. SPSS for further analysis and presentation. Finally it is possible to build up a general database with endoscopic results of different hospitals.


22

Material consumptionFor management of the endoscopy room it is possible to register the consumption of all the materials that are used during the endoscopy. Costs per endoscopy can be calculated and stock management can be performed.

The system can be connected to the disinfecting machines (ETD, Olympus) of the endoscopes to control and record the cleaning and disinfecting program of the endoscopes. Through its unique number, the endoscope that is used during the ex-amination is recorded and followed in the cleaning process. After the cleaning pro-cess endoscopes are dried by a constant flow of air. The correct drying process and storage of the scope is recorded as well and a message is shown if anything in the cleaning, disinfection and drying process went wrong. Also the maintenance and any repair of endoscopes can be documented in the system, making endoscopes traceable at any time. By this procedure the quality control of the endoscopy unit is assured.

Management of endoscopy unitDuring daily work with an Endoscopic Electronic Medical Record multiple endo-scopic data are stored. With a module named EndoQlik reports of all cost-related and logistical details are available whenever needed. These reports can improve transparency of costs and visualize any logistical problems.

Ease of useFor the successful implementation of a endoscopy recording and management sys-tem it is essential that the system is uncomplicated and user friendly. Any doc-tor, also the inexperienced computer user, must be able to work easily with a new program. The learning curve has to be relatively short as most doctors simply don’t have the time or interest to take a course just for using a program.

Secondly, it has to be a stable and fast working program. Any often-appearing error or fault in the program will lead to abandoning the program.

Thirdly, the program should be developed for the future and regularly updated with new features. The program should have options that can be used in the future but are not necessary at this moment. From the moment that a computerized pro-gram is in use, the users prefer to combine more tasks within the same program and to use only one computer program to manage the entire endoscopy unit.

Objectives and outline of the thesis

Gastrointestinal endoscopy investigations are an important part in examination and treatment of patients with complaints of the gastrointestinal tract, but also in

Chap ter 1

23

screening for malignancies and therapeutic interventions. The result of the visual investigations has to be formulated into a report by the endoscopist for the refer-ring physician and other professionals involved in the treatment of the patient. The quality of this report is of extreme importance for further management of the patient.

To enhance the quality of the reports and to perform research on gastrointes-tinal endoscopic data a specific standardized way of reporting was developed. The aim of this thesis was to analyze different aspects of this endoscopic electronic medical record system for report writing and stimulate further research with the produced data.

For composing an endoscopy report three different ways of report writing were created as mentioned above. In Chapter 2 the first developed two methods of stan-dardized report writing, that is standard reports and text blocks are described and discussed.

An often heard criticism on computerized reporting systems, especially in med-ical settings, is the time aspect required to work with them. Physicians often insist to keep their usual way of working. We therefore studied in Chapter 3 time aspects of three different ways of report writing. Computerized reports were compared to dictated and handwritten reports. The time needed to compose the report and to send it to the referring doctor was compared.

Starting with computerized systems demands a great investment in hardware, software and various links between the system and the hospital information sys-tem. Computerized systems are also known to record investigations more accurate, leading to billing benefits, and to improve quality aspects. In Chapter 4 a study on costs for the above mentioned three different ways of report writing (computerized, dictated and handwritten) to compose the report was performed. A cost benefit analysis was executed between dictated and computerized reporting.

In the computerized system, besides standard reports and text blocks also struc-tured data entry is allowed. For every examination the endoscopist is free to choose one of the three ways of report writing. To be able to compare the results of the investigations, the reports need to be coded in a similar way. For this reason a cod-ing system had to be developed. It was chosen to extend a widely used coding sys-tem, the International Classification of Diseases, 10th version, of the World Health Organization (WHO). In Chapter 5 this extension is described with a new developed chapter for endoscopic interventions. The complete coding system is available on www.trans-it.org.

The automatically recorded codes are stored in the database of each hospital. Within the Trans.it Project research will be performed on the anonymous data. Therefore a central database is created with the anonymous data of all investiga-tions in the participating hospitals. Within the project it was agreed to use stan-


24

dard reports and text blocks although the report maker can use free text or adjust the report. Obviously, this could potentially influence the correctness of the auto-mated terminology coding. In Chapter 6 a multicentre study is described to assess to what extent automatically generated GET-C codes are in agreement with the ac-tual text of the endoscopy report.

In the Ikazia Hospital, Rotterdam, the computerized system was used from 1996 onwards. It was possible to analyze the endoscopic data over a ten year period. Trends in different diagnoses were analyzed. In Chapter 7 gastric and duodenal ulcer disease is described over time. Classification of severity and re-bleeding risk are changing over time. In Chapter 8 different referring groups and their yield of upper gastrointestinal endoscopy is compared in an open access endoscopy unit.

Finally, in Chapter 9 the results of these studies are summarized and discussed.

Chap ter 1

25

References

1 Hirschowitz B. Endoscopic examination of the stomach and duodenal cap with the fiber-scope. Lancet. 1961;1(7186):1074.

2 Hirschowitz B. A fibre optic flexible oesophagoscope. Lancet. 1963;2(7304):388.3 Classen M. Back to the future: the first papillotomy at Erlangen. Gastrointest Endosc. 2000

May;51(5):637.4 Elewaut A, Cremer M. The History of Gastrointestinal Endoscopy - the European Perspec-

tive In: Classen M, Lightdale C, Tytgat G, editors. Gastroenterological Endoscopy. Stutt-gart: Thieme; 2002. p. 17-30.

5 Crespi M, Delvaux M, Schaprio M, Venables C, Zwiebel F. Working Party Report by the Committee for Minimal Standards of Terminology and Documentation in Digestive En-doscopy of the European Society of Gastrointestinal Endoscopy. Minimal standard termi-nology for a computerized endoscopic database. Ad hoc Task Force of the Committee. Am J Gastroenterol. 1996 Feb;91(2):191-216.

6 Shapiro M. Computerization of Endoscopic Reports. Endoscopy. 1992;24:478-80.7 Basisset Kwaliteitsindicatoren Ziekenhuizen. In: (IGZ) HCI, editor.2010.8 Faigel DO, Pike IM, Baron TH, Chak A, Cohen J, Deal SE, et al. Quality indicators for gas-

trointestinal endoscopic procedures: an introduction. Gastrointest Endosc. 2006 Apr;63(4 Suppl):S3-9.

9 Lieberman D, Nadel M, Smith RA, Atkin W, Duggirala SB, Fletcher R, et al. Standardized colonoscopy reporting and data system: report of the Quality Assurance Task Group of the National Colorectal Cancer Roundtable. Gastrointest Endosc. 2007 May;65(6):757-66.

10 Quality improvement of gastrointestinal endoscopy: guidelines for clinical application. From the ASGE. American Society for Gastrointestinal Endoscopy. Gastrointest Endosc. 1999 Jun;49(6):842-4.

11 Ariotti G AG, Ballerini P, Corti E. Micro-ordinateur et fichier de protocoles d’endoscopie: coût et benefice d’un systeme de base de données. Acta Endoscopica. 1985;15(419-429).

12 Conway JD, Adler DG, Diehl DL, Farraye FA, Kantsevoy SV, Kwon R, et al. Endoscopic elec-tronic medical record systems. Gastrointest Endosc. 2008 Apr;67(4):590-4.

13 Cooper G, Sivak MV, Jr. More on endoscopy databases. Gastrointest Endosc. 2000 Jun;51(6): 759-61.

14 Feczko PJ, Ackerman LV, Halpert RD, Simms SM. A computer-based gastrointestinal infor-mation management system. Radiology. 1984 Aug;152(2):297-300.

15 Gouveia-Oliveira A, Raposo VD, Azevedo AP, Salgado NC, Almeida I, Silva AM et al. SI-SCOPE: a multiuser information system for gastrointestinal endoscopy. Endoscopy. 1991;23:272-7.

16 Gouveia-Oliveira A, Salgado N, Azevedo A, Lopes L, Raposo V, Almeida I, et al. A unified approach to the design of clinical reporting systems. Methods of information in medicine. 1994;33(5):479.

17 Kruss D. The ASGE database: computers in the endoscopy unit. Endosc Rev. 1987;4:64-70.18 Reichel W. Computer systems in gastroenterologic endoscopy. An analysis of current avail-

ability on the Austrian market. Z Gastroenterol. 1993;31:183-91.19 Rozen P. Computer assistance in gastroenterology: an update. Am J Gastroenterol.

1988;83:1323-7.20 Stöltzing H, Birkner B, Lindlar R, Ohmann C, Zaczyk R, Kuntzen O, et al. Computer-as-

sisted documentation in upper gastrointestinal endoscopy: experiences with routine use at 3 clinics. Zeitschrift für Gastroenterologie. 1989;27(11):667.


26

21 Waye JD AL, Alvarez S, Archila P, Bornman PC, Hashiba K et al. Endoscopy reports, data-bases, and computers in 2001. Gastrointest Endosc. 2001;53:838-9.

22 Westerheim J AL, Hofstad B, Larsen S, Osnes M. The SADE system: an endoscopic database manager. Endoscopy. 1991;23:265-8.

23 Zimmerman J, Shapiro E, Porath Y, Rachmilewitz D. Computer application in a gastro-intestinal endoscopy unit. Endoscopy. 1985;17(6):207-9.

24 Health Level 7 international. Available from: http://www.hl7.org.25 Groenen MJM, Th. Ouwendijk RJ, Middelkoop MPC, Dees A, Bac DJ, van Berge Henegou-

wen GP. The yield of open access upper gastrointestinal endoscopy. Evaluating 5943 upper endoscopies using Endobase - a standardised report system. European Journal of Gastro-enterology & Hepatology. 1999;11(12):A92.

26 Delvaux M, Crespi M. the Computer Committee of ESGE. Minimal standard terminology in digestive endoscopy. Version 2.0. Endoscopy. 2000;32:159-88.

Chap ter 1

27

1 Department of Gastroenterology and Hepatology, Ikazia Hospital, Rotterdam

2 Department of Gastroenterology and Hepatology, Erasmus Medical Center, Rotterdam

3 Department of Gastroenterology and Hepatology, University Medical Center Utrecht

4 Department of Gastroenterology and Hepatology Academic Medical Center, Amsterdam

Computerization

of endoscopic reports

using standard reports

and text-blocks

Ne t hJ Med.2 0 0 6M a r ;6 4 ( 3): 7 8 - 8 3

MarcelJ.M.Groenen1,2

ErnstJ.Kuipers2

GerardP.vanBergeHenegouwen3

PaulFockens4

RobJ.Th.Ouwendijk1

Chap ter 2

29

Abstract

Background The widespread use of gastrointestinal endoscopy for diagnosis and treatment requires effective, standardized report systems. This need is further in-creased by the limited storage of images, and by the need for structured databases for surveillance and epidemiology. We therefore aimed to a report system which would be quick, easy to learn, and suitable for use in busy daily practice.

Methods Endobase III® is an endoscopic information system offering three differ-ent ways of report writing, i.e. standard reports, text-blocks and Minimal Standard Terminology (MST). A working group of two academic and four general hospitals worked as a reference group for development of standard reports and text-blocks. Guidelines of different gastrointestinal endoscopic societies were followed to com-pose the reports

Results Standard reports were based on a list of distinct diagnoses, text-blocks were based on anatomic landmarks and individual procedures. As such, 316 stan-dard reports were developed for upper and lower gastrointestinal endoscopy, and ERCP. In this way selecting one diagnosis produces a complete report. A total of 1571 different text-blocks were additionally developed for every part of the GI-tract and for procedures during endoscopy. This module allowed generation of a full report on combination of text blocks. Reports could be composed and printed within two minutes for 90% of cases.

Conclusions Standard reports and text-blocks is a quick, user-friendly way of re-port writing accepted and used by different gastroenterologists in the Netherlands.

2

30

Chap ter 2

Introduction

Gastrointestinal endoscopy has become a standard, widely available technique for diagnosis and treatment of gastrointestinal disorders. The number of endo-scopic procedures is ever increasing, among others as a result of the continuous development of newer techniques, the introduction of screening and surveillance programs for gastrointestinal disorders, and the increasing incidence of various gastrointestinal disorders. A recent survey among endoscopists in the Nether-lands showed that 325.000 gastrointestinal endoscopies are performed annually in a population of 16.000.000.1 As an imaging method with numerous repetitive maneuvers as well as findings, gastrointestinal endoscopy reports are particu-larly suitable for electronic storage and processing.2 Besides this there is a need for structured databases for surveillance, epidemiology, quality control and research. This need is further increased by the limited storage of images during endoscopy, making the report essential.

For that purpose, several endoscopic information systems have been developed in the past decade to record endoscopic findings, to store images and to compose reports.3-13

Most of these systems are stand alone report systems, not suitable for imple-mentation in a hospital information system. The combination of report writing and digital image storing is not available in all the systems. The structure of the database of most of the systems is poorly accessible for research and export of data.

There are several crucial criteria for a report system that need to be fulfilled to get it generally used, suitable for every hospital and implemented in a hospital information system.

In further development of healthcare informatics it is necessary for such sys-tems to be readily acceptable for most endoscopists in a hospital unit. Secondly it should be possible to exchange and compare data and digital images between dif-ferent consultants and hospitals. Standardized protocols should be used to com-municate between different systems within a hospital based on the Health Level 7 – protocol (HL-7). For exchanging images a standard format such as DICOM (Digital Imaging and Communications in Medicine) is essential.

To get the system accepted in daily practice it is crucial that first of all, data entry is fast and accurate. Thus the system has to be accessible for the computer illiterate and the learning time should be limited.14

Programs using the currently available structured data-entry, like the Minimal Standard Terminology (MST),15-18 do not fulfill all these crucial criteria. Firstly, com-position of a report by means of MST is usually time-consuming because of the dif-ferent options available. Secondly, there is a risk of getting lost in the data-entry module, caused by the numerous available choices that has to be selected.

31

Computerization of endoscopic reports

Our aim was to develop a report system that is quick, easy to learn and can be used in busy daily practice by any endoscopist. Moreover, we considered it necessary that the program would have the capacity to build up a database with endoscopic findings for various purposes including management of surveillance programs, and epidemiological studies and quality control. Therefore the findings should be linked to a specific comprehensive code-system. This would allow anonymous evaluation of data.

Finally, consensus of gastroenterologists from different hospitals should be achieved for use of the new report system.

Materials and Methods

Endoscopic Information SystemIn the latest version of Endobase III®, developed by Olympus Software, it is possible to combine different text-blocks to compose a complete report besides the use of standard reports and MST.

After selecting the different standard reports, text-blocks or MST the composed report can be adapted in a word processor. In the program an extensive relational database structure was build. In this way it became suitable to store all different data produced in an endoscopy unit, digital images and video included, and retract it separately with all kinds of queries. Also a structured data entry is available, the Minimal Standard Terminology (MST). The MST has been translated into Dutch in 1998 by our group in co-operation with Dr Delvaux during a workshop on MST.

Trans.it-working groupIn the end of 1999 a working group, the Trans.it-project group, has been founded as a peer reference group to design the standard reports and text-blocks that were developed and used in the endoscopy units by the participating gastroenterolo-gists. This group gathers on a regular basis to discuss the reports, a comprehensive coding system and new developments for endoscopic information systems.

The Trans.it working group consist of two academic hospitals and 4 general hospitals and perform about 15% of all gastrointestinal endoscopies in the Nether-lands. All the participants of the Trans.it-group use the same version of standard reports, text-blocks and translated MST. An alteration in the content of a standard report or text-block will only be executed with agreement of a majority of the work-ing group members.

All various standard reports and text-blocks are directly linked to a specific code. The codes are based on the ICD-10 code system and are extended for specific endo-scopic findings.19

32

Structure of an endoscopic reportSeveral committees of societies for Gastrointestinal Endoscopy have proposed guidelines to get a standardized format for endoscopic reports. Considering the American Society of GE proposal,20 the European Society of GE adaptation,15 the ad-vice of the Netherlands Society of Gastroenterology and our experiences with an electronic report-system we developed an extended structure of an endoscopic re-port suitable for our endoscopy units [Table 1].

For the description of the findings at the investigation a list of items proposed by the Netherlands Society was used [Table 2].

We used this structure and the proposed items as guidelines to compose the stan-dard reports and text-blocks in our system. The grading and severity of findings is classified by for example the LA-Classification for reflux disease21 and Forrest clas-sification22 for ulcers. Preceding the examination most of the basic data of the patient necessary for the endoscopic report, like indication, medication, endoscopist, endoscope iden-tification number, referring doctor, general practitioner, medical history and risk factors, are already recorded in the system. The patient data can be extracted with HL-7 protocol from the Hospital Information System by using the personal identi-

Chap ter 2

Table 1 Usedstructureofanendoscopicreport

• Patientidentificationdata• Dateofprocedure• Referringdoctor• Endoscopist• Assistingdoctor• Instrumentsused• Reasonsforexamination• Preparation• Typeofendoscopicexamination• IdentificationnumberoftheEndoscope• Medication(Anaesthesia,Analgesia,Sedation)• Anatomicalextentofexamination• Limitation(s)ofexamination• Findingsandspecimensobtained• Therapeuticintervention(s)andresult(s)• Notationofimagescaptured• Complications(duringendoscopyandwithin24-48hours)• Endoscopicdiagnosis• Recommendationsforreferringdoctor• Comments• Recallletter

33

fication number (PIN) of the patient. Other features are recorded during or shortly after the examination date e.g. Helicobacter Pylori tests, histology or laboratory results, complications appearing after the examination and results of other gastro-intestinal examinations like ultrasonography, X-ray or manometry studies.

Presentation and selection of different text-blocksThe presentation of the different standard reports and text-blocks was based on the experience that endoscopists translate their findings into a diagnosis at the end of an endoscopy. To shorten the time of searching the corresponding diagnosis, the text-blocks are presented in different subsections. First of all different text-blocks were divided into anatomical regions, that are easily defined during endoscopic in-vestigations, e.g. esophagus, stomach, duodenum. Within an anatomical region the possible different diagnosis are grouped, e.g. esophagitis contains reflux, caus-tic, viral. Within these groups eventually a classification or grading is added.

All the text-blocks are presented alphabetically in the program. By typing the first characters of a diagnosis the selection of the group of diagnosis is presented.

Results

Standard reportsBased on individual diagnoses, we constructed 316 different standard reports. One-hundred-thirty-four of these reports pertained to esophago-gastro-duodenoscopy, 143 to lower digestive endoscopy, and 39 to ERCP. In an open-access endoscopy unit


Table 2 ItemsusedtodescribefindingsatupperGIendoscopy

• Usemmorcmindescribingthedimensionsofalesion• Findingsinoesophagus.Givedistanceincmfromlesionstoteeth• DistanceofZ-linetoteeth• Distanceofhiatalnarrowingtoteeth• Aspectofcontentsofstomach• Peristalticandinflationofthestomach• Findingsinantralregion• Findingsincorpusofthestomach• Findingsincardiaandfundusinretroversion• Findingsinangularregion• Findingsinpylorusandpassing• Findingsinduonalbulb• Findingsinproximalduodenum• Locationofbiopsiestaken• Captureofimages• Otherprocedures• Commentonproceedingsofexamination

34

at a district general hospital in 32,3% of the endoscopic examinations no abnormali-ties are found during endoscopy.23 Likewise in our own data these same numbers of around 30 % are found between two different referring groups.24 The reports com-posed for these examinations are simple and fully standardized. Nevertheless all the items listed in table 2 have to be included, to obtain a complete report.

The reports of the remaining 67,7% of the endoscopic examinations, where at least one abnormality is found, must also contain all items to make them com-plete. In some of these examinations only one abnormality is found leaving the re-mainder of the examination without any abnormalities. These examinations can also be reported using standard reports.

Chap ter 2

Table 3 Examplesofsomedifferentstandardreportsforoesophago-gastro-duodenoscopy

Barrett’smucosa 1 Barrett’smucosa2 Barrett’smucosawithrefluxoesophagitisgradeA3 Barrett’smucosawithrefluxoesophagitisgradeB4 Barrett’smucosawithrefluxoesophagitisgradeC5 Barrett’smucosawithrefluxoesophagitisgradeD6 Barrett’smucosacontrolendoscopy7 Barrettcarcinoma

Refluxesophagitis 8 RefluxoesophagitisgradeA9 RefluxesophagitisgradeB10 RefluxesophagitisgradeC11 RefluxesophagitisgradeD12 RefluxesophagitisgradeDwithulcer13 RefluxesophagitisgradeDwithstricture

DuodenalUlcer* 14 Duodenalulcer,Spurtingbleeding(ForrestIa)15 Duodenalulcer,Non-spurtingactivebleeding(ForrestIb)16 Duodenalulcer,Visiblevessel,noactivebleeding(ForrestIIa)17 Duodenalulcer,Non-bleedingwithoverlyingclot(ForrestIIb).18 Duodenalulcer,Withhematin-coveredbasis(ForrestIIc)19 Duodenalulcer,Cleanulcerground,noclot,novessel(ForrestIII)20 Normaloesophago-gastro-duodenoscopy

Hiatalhernia 21 Slidinghiatalhernia22 SlidinghiatalherniawithCameronlesions23 Slidinghiatalherniaandgastritis24 Slidinghiatalherniaandgastritisandduodenitis

Varices 25 VaricesoesophagusgradeI26 VaricesoesophagusgradeII27 VaricesoesophagusgradeIII28 VaricesoesophagusgradeIV29 Varicesbleedingbanding30 Varicesbleedinginjection

*Similarstandardreportsforgastriculcer

35

Other examinations show more abnormalities, making standard reporting less ap-plicable. In less common combinations of abnormal findings or rare findings, the use of the specific text-blocks is recommended.

The composed standard reports are based on the endoscopic diagnoses or a com-bination of diagnoses made during endoscopy [Table 3]. After the examination the endoscopist has to select this endoscopic diagnosis out of the list of different stan-dard reports.

For reflux esophagitis six different standard reports are created for the generally used LA-classification, grade A to D and an ulcer or stricture of the esophagus. For the frequently seen combination of columnar mucosa (Barrett) and reflux esophagi-tis four additional standard reports with this combination are written.

Gastric and duodenal ulcers are described according to the Forrest classification, resulting in 30 different standard reports for a number of various locations.

Infrequent findings or findings at rare locations can be described with the use of text-blocks.

The reports are alphabetically arranged in Endobase and can be searched for by giving the first one or more characters of the diagnosis.

During a normal program at our endoscopy unit the time needed to compose a report by selecting standard reports was measured. Various endoscopists composed a total of 291 reports in this way. A student was positioned behind the endoscop-ist and timed different items during report writing. The average reporting time including selection of the standard report, addition of some details in the word processor and printing of the report was 1 minute 21 seconds (SD 51 seconds) for stan-dard reports.

Text-blocksThe text-blocks were divided in different sections and presented in tabs according to different anatomical sections seen during the endoscopy and some specific parts. Reports created with text-blocks were composed by selecting a diagnosis or finding from different sections of text-blocks.

For upper endoscopy 8 different sections were made [Table 4] [Image 1].

First of all the preparation and progress of the examination was selected. Four sec-tions were designed for the various anatomical regions: esophagus, stomach, duo-denal bulb and descending duodenum. A separate section was made for aberrant anatomy after gastrointestinal surgery. One section consisting of ‘therapeutic’ in-terventions, e.g. taking of biopsies and placing of endoprothesis. Another section was composed with different kinds of recommendations to the referring doctor. Finally a section with conclusions which is automatically built up by the different selected text-blocks.


36

It is possible to select one or more text-blocks from each section, but also to select none and omit a section. A total of 252 text-blocks were created for upper endoscopy.

Lower gastrointestinal endoscopy exists of 13 different sections [Table 4]. Again it starts with the preparation and progress of the examination. Seven different sections for anatomical regions: Ileum, cecum, ascending colon, transverse colon, descending colon, sigmoid colon, rectal and anal region. A separate section was made for digital rectal examination. Again one section for post surgery anatomy and a section for therapeutic interventions. Recommendations for the referral doc-tor are within the last section. A total of 607 text-blocks were created for lower en-doscopy.

Reports of endoscopic retrograde cholangiopancreatography (ERCP) were main-ly composed with text-blocks and consist of different anatomical and therapeutical parts [Table 4].

Each text-block consists of one or more sentences describing a diagnosis or find-ing of that particular text-block. For this purpose a total of 1571 different text-blocks were written and used at this moment [Table 5].

Chap ter 2

Figure 1 Text-blockswithdifferentsections

37

In the same way as with using standard reports the time needed to compose a re-port with text-blocks was timed. In total 133 examinations were reported by differ-ent endoscopists and the needed time was measured. The mean time for selecting different text-blocks, making some adaptations in the word processor and printing of the report was 1 minute 37 seconds (SD 55 seconds).


Table 4 Differentsectionsfortext-blocks

Examination

Oesophagogastroduodenoscopy Colonoscopy ERCP

Preparationandprogressofexamination Digitalrectalexamination IntroductionandPreceeding

Oesophagus Preparationandprogressofexamination

PapillaMajor

Stomach Ileum PapillaMinor

Duodenalbulb Cecum CanulationandPre-Cut

Descendingduodenum Ascendingcolon Commonbileduct

Post-surgery Transversecolon Cysticductandgallbladder

Therapeuticinterventions Descendingcolon Bifurcationandhepaticducts

Conclusions Sigmoidcolon Pancreaticduct

Advice Rectalandanalregion Sphincterotomyandballoondilatation

Post-surgery Therapeuticinterventionsbileduct

Therapeuticinterventions Therapeuticinterventionspancreaticduct

Conclusions Conclusions

Advice Advice

Table 5 NumberofStandardreportsandText-blocks

Examination Number of Standard Reports Number of text-blocks

Oesophago-gastro-duodenoscopy 134 252

Sigmoidoscopy 47 420

Colonoscopy 96 697

ERCP 39 202

Total 316 1571

38

In comparison the using of MST was also measured in 250 reports made by an ex-perienced user. The mean time for this way of report writing is 2 minutes and 50 seconds (SD 1 minute 10 seconds).

CodingAll endoscopic reports are coded automatically with an extension of the ICD-10. The different used report systems all produce the same code for identical findings. In this way extensive research possibilities are created. For example a search on 13081 upper endoscopies for a specific ICD-10 code for duodenal ulcers (K26) result in 511 (3.9%) duodenal ulcers. In time this incidence declines from 4.1% in 1996 to 2.8% in 2005. Twenty-one of these duodenal ulcers (4.1%) showed active bleeding and were classified as Forrest I, six of these as Forrest Ia. Signs of recent bleeding were found with a visible vessel in 45 patients (8.8%), an overlying clot, Forrest IIb, in 31 (6.1%) and a haematin covered basis, Forrest IIc, in 28 patients (5.5%).

DiscussionsStructured computerized report systems are essential for modern gastrointestinal practice. They should enable systematic, rapid, informative, comprehensive report-ing of endoscopic findings and at the same time allow database handling for vari-ous purposes. Potentially, they should also be used for safety and quality control, as well as other issues including maintenance of equipment, management of stocks, and billing. We showed in this study that a structure report system, in our setting the Olympus developed Endobase III® system, allows incorporation of standard re-ports as well as text blocks. With 316 reports and 1571 text blocks, 90% of the endo-scopic examinations could regularly be composed within 2 minutes. This makes it useful for the busy daily practice of many endoscopy units. All endoscopists in the participating hospitals use this system for report writing in every case. Standard reports can be used to report examinations without abnormalities and examinations with frequently seen abnormalities. In case of more rare findings and/or a combination of diagnoses the use of text-blocks is more suitable. This still makes it possible to compose a comprehensive report of the performed examina-tions in a short time. For those examinations (ca 5-10 %) where it is hard to compose a report with standard reports or text-blocks, we propose to use a standard struc-tured data entry like the MST. In our experience MST is more complex, takes more time and there is a risk of getting lost in the data tree. The advantage is that you can describe the findings in a very punctual way and build up a structured database.

In comparison, in using MST to compose a complete report you have to make about 40 different choices for the description of an examination with only a few ab-normalities. The possibility for the endoscopist to choose the type of report writing after the examination makes the program user-friendly and well accepted. With

Chap ter 2

39

standard reports and text-blocks it is possible to register a standard list of all the requirements on medical records and endoscopic reports in particular.25 With all legal consequences nowadays, registration of endoscopic information should be as complete as possible. With this system all this information can be stored and easily retrieved.

All the standard reports and text-blocks are directly linked to an extended ICD-10 code system in the database. Also other data in the system like reason of examina-tion, medication, complications are coded. With these codes an anonymous data-base can be build with endoscopic data from different hospitals.

The Standard reports and text-blocks are written in Dutch, and will be trans-lated. They are used in the Endobase system, but can be applied to any system that can work with text-blocks and a code system.

All the reports and text-blocks are tested and eventually adapted by the Trans.it working group. The Trans.it working group will stay operational for at least three years, in order to improve the functionality and quality of the reports and to create a large anonymous central database. After three years we will have the possibility to answer specific research questions from the results of a database with approxi-mately 60.000 upper endoscopies performed in a uniform way.

Nowadays the system with the standard reports and text-blocks is accepted and used in about 30 % of the Dutch Hospitals.


AppendixMembers of the Trans.it Project group includes, in alphabetical order:G.P. van Berge Henegouwen, M. Bruno, P.J. Bus,J.A.G. Drapers, P. Fockens, M.J.M. Groenen, M. Grubben,G. den Hartog, G.H.J. van der Hoorn, P.J. Kingma, A.W.M. van Milligen de Wit, S.A. Mulder, P. Niermeijer, R.J.Th. Ouwendijk, P.J. van der Schaar, T. Schwartz, R. Soekhoe, W.N.H.M. Stuifbergen, A.A. Tanis, F.P. Vleggaar,P.J. Wahab.

40

41

References

1 Kolkman JJ, Bartelsman J, Mulder C, Mestrom T. Nederland: per jaar 325.000 endoscopieën (Dutch). Magma. 2000;4:5-7.

2 Classen M, Wagner F, Swobodnik W. Electronic data base in gastroenterological endos-copy. Endoscopy. 1991 Jan;23(1):29-31.

3 Ariotti G, Allosia GP, Ballerini P, Corti E. Micro-ordinateur et fichier de protocoles d’endoscopie: coût et benefice d’un systeme de base de données. Acta Endoscopica. 1985;15(5):419-29.

4 Cooper G, Sivak MV, Jr. More on endoscopy databases. Gastrointest Endosc. 2000 Jun;51(6): 759-61.

5 Feczko PJ, Ackerman LV, Halpert RD, Simms SM. A computer-based gastrointestinal infor-mation management system. Radiology. 1984;152(2):297-300.

6 Gouveia-Oliveira A, Raposo VD, Azevedo AP, Salgado NC, Almeida I, Silva AM, et al. SI-SCOPE: a multiuser information system for gastrointestinal endoscopy. Endoscopy. 1991;23(5):272-7.

7 Kruss DM. The ASGE database: computers in the endoscopy unit. Endosc Rev. 1987;4:64-70.8 Reichel W. [Computer systems in gastroenterologic endoscopy. An analysis of current

availability on the Austrian market]. ZGastroenterol. 1993;31(3):183-91.9 Rozen P. Computer assistance in gastroenterology: an update. AmJGastroenterol.

1988;83(12):1323-7.10 Stoltzing H, Birkner B, Lindlar R, Ohmann C, Zaczyk R, Kuntzen O, et al. [Computer-

assisted documentation in upper gastrointestinal endoscopy: experiences with routine use at 3 clinics]. ZGastroenterol. 1989;27(11):667-75.

11 Waye JD, Aabakken L, Alvarez S, Archila P, Bornman PC, Hashiba K, et al. Endoscopy reports, databases, and computers in 2001. Gastrointest Endosc. 2001 Jun;53(7):838-9.

12 Westerheim J, Aabakken L, Hofstad B, Larsen S, Osnes M. The SADE system: an endoscopic database manager. Endoscopy. 1991;23(5):265-8.

13 Zimmerman J, Shapiro E, Porath Y, Rachmilewitz D. Computer application in a gastroin-testinal endoscopy unit. Endoscopy. 1985;17(6):207-9.

14 Gouveia-Oliveira A, Salgado NC, Azevedo AP, Lopes L, Raposo VD, Almeida I, et al. A uni-fied approach to the design of clinical reporting systems. Methods Inf Med. 1994 Dec;33(5): 479-87.

15 Crespi M, Delvaux M, Schapiro M, Venables C, Zwiebel F. Minimal standards for a comput-erized endoscopic database. Ad hoc Task Force of the Committee for Minimal Standards of Digestive Endoscopy of the European Society of Gastrointestinal Endoscopy (ESGE). Am J Gastroenterol. 1994 Aug;89(8 Suppl):S144-53.

16 Crespi M, Delvaux M, Schaprio M, Venables C, Zwiebel F. Working Party Report by the Committee for Minimal Standards of Terminology and Documentation in Digestive Endoscopy of the European Society of Gastrointestinal Endoscopy. Minimal standard ter-minology for a computerized endoscopic database. Ad hoc Task Force of the Committee. AmJGastroenterol. 1996;91(2):191-216.

17 Delvaux M, Korman LY, Armengol-Miro JR, Crespi M, Cass O, Hagenmuller F, et al. The minimal standard terminology for digestive endoscopy: introduction to structured report-ing. IntJMedInf. 1998;48(1-3):217-25.

18 Logan JR, Klopfer KC. The use of a standardized terminology for comparison of free text and structured data entry. Proc AMIA symp. 2000:512-6.


42

Chap ter 2

19 Groenen MJM, Ouwendijk RJT, van Berge Henegouwen GP. Gastrointestinal Endoscopic Terminologie Coding. GET-C; an extension of ICD-10. Endoscopy. 1999;31 (suppl 1):E92: P1103E-.

20 Shapiro M. Computerization of Endoscopic Reports. Endoscopy. 1992;24(Suppl 2):478-80.21 Lundell LR, Dent J, Bennett JR, Blum AL, Armstrong D, Galmiche JP, et al. Endoscopic

assessment of oesophagitis: clinical and functional correlates and further validation of the Los Angeles classification. Gut. 1999;45(2):172-80.

22 Forrest JA, Finlayson ND, Shearman DJ. Endoscopy in gastrointestinal bleeding. Lancet. 1974;2(7877):394-7.

23 Boulton-Jones JR, Follows MC, Mahmoud AA. Open-access endoscopy: are age-based guidelines justified? An audit of experience of 1000 open-access endoscopies at a district general hospital. Endoscopy. 2003;35:68-73.

24 Groenen MJM, Ouwendijk RJT, Middelkoop MPC, Dees A, Bac DJ, van Berge Henegou-wen GP. The yield of open access upper gastrointestinal endoscopy. Evaluating 5943 upper endoscopies using Endobase a standardised reportsystem. European Journal of Gastroen-terology & Hepatology. 1999;12:Abstract

25 Korman LY. Standardization in endoscopic reporting: implications for clinical practice and research. JClinGastroenterol. 1999;28(3):217-23.

43



3 Department of Medical Informatics, Erasmus Medical Center, Rotterdam,

4 Department of Gastroenterology and Hepatology, Albert Schweitzer Hospital, Dordrecht

5 Department of Gastroenterology and Hepatology, Sint Franciscus Gasthuis, Rotterdam

Computerized endoscopic reporting

is time competitive

with conventional methods

EurJ In t er nMed.2 0 0 7 Jul ; 1 8 (4):3 2 1 - 5

JagdieshK.Soekhoe1


AstridM.vanGinneken3

G.Khaliq1

WilcoLesterhuis4

AntonieJ.P.vanTilburg5

RobJ.Th.Ouwendijk1

Chap ter 3

45

Abstract

Background Endoscopists use different methods for reporting a gastrointestinal endoscopy. These may result in handwritten, dictated or computerized reports. The time needed to create the report is an important parameter for acceptance of the used method. Besides the issue of time-consumption it is important to be aware of the possible advantages and disadvantages of these different methods.

Aim of the study was to compare time aspects of different methods of report writing.

Methods Three different methods of report writing, i.e. handwritten, dictated and computerized were compared. In three different endoscopy departments one inves-tigator recorded the time needed to compose the report and send it to the referring doctor.

The time needed to describe different diagnoses at endoscopy was compared be-tween the systems.

Results Handwritten reports were completed in an average time of 113 seconds, free-text dictated reports by the endoscopist in 65 seconds where additional time for the typist was 172 seconds, and computerized predefined reports in 86 seconds. The incidences of abnormalities found in the reports of the different hospitals were comparable.

Conclusion To a large extent, computerized predefined reports could be composed in almost the same time as handwritten and dictated reports. Free-text dictated and computerized predefined reports are both stored in the hospital information system, but only computerized predefined reports including endoscopic pictures are stored in a structured database, which makes statistical analysis possible.

3

46

Chap ter 3

Introduction

Approximately 400,000 endoscopies are carried out in the Netherlands each year. Endoscopy reports are extremely important to inform the referring doctor about the findings of the investigation. At present, endoscopy results are recorded in dif-ferent ways namely free-text handwritten reports, free-text dictated and typed reports or computer-generated standardized reports using databases with either predefined text or structured data entry.

Computer systems have become a part of everyday working life in all areas of work, including hospitals. The aim of using computer technology is usually to im prove quality, reduce costs, save time, collect research data and also for conve-nience. Although several commercial reporting programs are available, they are not widely used.1-2 Some of these systems are not found sufficiently satisfactory be-cause of their complexity of use and the time needed to create a report.3-4

The publication of several national and international recommendations regard-ing the documentation of findings and images, represents evidence of a growing demand for quality control among endoscopists and other health care providers.5-6 In the mean time, advances in information technology and software engineering have yielded more powerful tools for reporting. Computerized reporting may pro-mote doctors’ acceptance of endoscopy standards, improve the quality of the report, and include topics that tend to be omitted in free text reports.7-10

Computerized reporting may have advantages compared to other report writing methods, but might be more time-consuming. To establish whether time is the limiting factor for computerized reports, existing methods of report writing were compared.

Materials and Methods

The study was carried out in three medium-size hospitals that used three different report writing systems for endoscopy results, i.e. handwritten, dictated free-text, and computerized reports based on predefined text blocks. In a period of six months one investigator visited the endoscopy units of the study hospitals. Reporting times of endoscopies, performed as part of the regular endoscopy program were evalu-ated. All upper and lower gastrointestinal endoscopies were included in a defined period. Upper or lower gastrointestinal (GI) endoscopies were evaluated separately during the study. Total report processing time was recorded by using a stopwatch and was noted in seconds. The main diagnosis of each endoscopy was written down. A total of 217 endoscopies were evaluated in the form of hand-written reports, 201 as dictated reports, and 232 as computerized predefined reports. Data were analysed

47

Time aspects of computerized endoscopic reporting

in Statistical Package for the Social Sciences (SPSS), version 11.5 (SPPS, Chicago, IL). Data are described by means and 95%-confidence interval (CI).

Handwritten reportsTwo experienced endoscopists participated in collecting the data in this hospital. The endoscopists made handwritten reports straight after the endoscopy. The en-doscopists wrote the reports in free text on a blank form, on which the secretary printed the patient’s identity through a punched card. The reports were handed over to the patient or sent to the referring doctor and a copy was stored and filed in the endoscopy unit.

When necessary, photographs were printed using a photo printer. Neither reports nor photos where saved in an electronic database. Reporting time was recorded and started upon beginning with writing the report and ended when the endoscopist finished the report. Photos were only printed in a minority of the investigations and therefore capturing and printing time was not recorded.

Dictated reportsTwo experienced endoscopists participated in collecting the data in this hospital. Dictated reports were partly created by the endoscopist and partly by a typist. After an endoscopy the endoscopist dictated the findings in free-text on a voice recorder. Immediately afterwards the typist typed out the report in a text editor, which was linked to the hospital information system and the computer diary of the endoscopy department. From this diary the patient’s identity, the name of the referring doc-tor, and the type of examination were retrieved. The verbatim text was typed out without using fixed text blocks. In case the general practitioner ordered the inves-tigation, the report was handed over to the patient straight away. In other cases reports were sent to the referring doctor by mail. All the reports were stored in the hospital information system, making them accessible from any computer in the hospital at any time.

In analogy to handwritten reports, captured images were printed with a photo printer. Time recording was divided in two parts, dictating time by the endoscopist and typing time by the typist. Again printing of photos was not included in the recorded time because it was only performed in a few investigations.

Computerized reportsTwo experienced endoscopists participated in collecting data in this hospital. Com-puterized predefined reports were created by the endoscopists with a commercially available software system, called Endobase III® of Olympus Software. Patient infor-mation was retrieved from the hospital information system. The type of examina-tion, the referring doctor and the endoscopist planned to perform the procedure,

48

were retrieved from the computer diary. The endoscopists entered through a pre-defined selection list details such as indication of endoscopy and medication. Re-ports could be created in two different ways, by predefined ‘Standard Reports’ or ‘Text Blocks’ (11). Standard reports describe an entire endoscopy without or with one abnormality. Text blocks describe only one part of the gastrointestinal tract with a specific abnormality. All standard reports and text blocks are linked to one or more specific gastrointestinal endoscopic terminology codes (GET-C) based on the International Classification of Diseases (ICD-10) (12). Images could be unlimited captured during endoscopy by a remote button on the endoscopes and saved in the database. Afterwards up to three images were selected and printed in the report on plain paper. The final text of the report could be edited, e.g. entering free text or filling in the measurements like the size of an abnormality. The reports were authorized, printed and automatically saved in the computer database and the hos-pital information system, making them accessible from any computer in the hospi-tal at any time. Reports were sent to the referring doctor immediately after endos-copy or given to the patients to hand them over to their general practitioner. Time for computer recording was recorded as follows: time for selecting photographs to be printed in the report, time for creating the report by selecting standard reports or text blocks and time for editing the report and free-text. These are consecutive actions.

Photo selection is a standard part of computerized reporting. Photo selecting time was recorded separately for computerized reports to make a reliable compari-son of the net reporting time compared to handwritten and dictated reports.

Results

A total of 650 endoscopy reports were evaluated. Respectively 217 handwritten re-ports (105 upper and 112 lower GI reports), 201 dictated reports (106 upper and 95 lower GI reports) and 232 computerized predefined reports (102 upper and 130 lower

Chap ter 3

Table 1 Meantimeinsecondsforendoscopistsinvolvedincomposingreport

Handwritten reports Dictated reports Computerized reports

UpperGI 114(109-120) 69(64-75) 74(65-84)

LowerGI 113(108-117) 61(55-67) 95(86-104)

Allendoscopies 113(110-117) 65(61-69) 86(79-93)

Meantimeinseconds(95%confidenceinterval)forendoscopistsinvolvedindifferentmethodsforupperlowergas-trointestinal(GI)endoscopiesandallendoscopieswithoutselectingendoscopicimages.

49

GI reports) were evaluated. The mean time the endoscopists of the three hospitals were involved in composing the report is shown in the Table 1.

Handwritten reports were performed in a mean time of 114 seconds (CI 109-120) for upper and 113 seconds (CI 108-117) for lower GI endoscopies respectively.

Dictated reports were completed in 242 seconds (CI 225-258) for upper and 232 sec-onds (CI 213-251) for lower GI endoscopies respectively. The time the doctor needed to dictate the report was 69 seconds (CI 64-75) for upper and 61 seconds (CI 55-67) for lower GI endoscopies respectively. The remaining time was needed to type the re-port by the typist, which took a mean time of 172 seconds (CI 163-181).

A report based on computerized predefined text blocks took a mean time of 74 seconds (CI 65-84) for upper and 95 seconds (CI 86-104) for lower GI endoscopies re-spectively. An additional mean time of 20 seconds (CI 19-22) was needed to select the endoscopic images

For the different methods the mean times of all endoscopies, divided in differ-ent stages, are shown in detail in Table 2. As also the diagnoses were recorded, it was possible to compare the average re-port times for the different diagnoses for each report method. These are presented in Figure 1 for upper and Figure 2 for lower gastrointestinal endoscopies.

In about 25-35% of the endoscopies performed no abnormalities were found. It is no-ticeable that normal findings and most often seen abnormalities could be described faster or in similar time by computerized report writing. Furthermore, in all three types of endoscopy reporting, composing a report took longer if the diagnosis was more complex.


Table 2 Meantotaltimeforcomposingcompletereports

Writing Dictating Typing SelectingPhotos

Composing Report

Total

HandwrittenN=217

113(108-117) 113(108-117)

DictatedN=201

65(61-69) 172(163-181) 237(225-250)

ComputerizedN=232

20(19-22) 86(79-93) 102(95-110)

Meantimeinseconds(95%confidenceinterval)forcomposingcompletereportinseconds.N=numberofreportsencounteredinthisstudy.‘Selectingphotos’istimeneededtoselectphotostobeprintedinthereport.‘Composingreport’istimetoselectapredefinedstandardreportortextblocksandeditingthereportcompletingifdesiredwithfreetext.‘Writing’isthetimeneededtohandwritethereport.‘Dictating’isthetimefortheendoscopisttodictatethereportontape.‘Typing’isthetimeneededforthetypisttotypeandprintthereport.

50

Discussion

This study shows that preparing an endoscopy report using a computer-generated method is not more time consuming than the conventional way. Contrary to the 5 minutes that have been suggested by Waye et al,4 computerized predefined reports including the selection of endoscopic images can be performed in 102 seconds (CI 95-110). These reports are produced significantly faster than dictated reports and almost equalling handwritten reports [Table 2]. It should be noted that the actual time an endoscopist is involved in creating a dictated report is lower than the time creating a computerized report [Table 1]. However, if we include report checking in the time necessary to create the report, an additional 10–30 seconds are needed to finalize and approve the dictated report. Adding an average of 20 seconds to dicta-tion time would result in an average time of about 85 seconds the endoscopist is en-

Chap ter 3

Figure 1 Endoscopistinvolvedtimeforuppergastrointestinalendoscopies,groupedbydiagnosis

0

50

100

150

200

Refluxdisease OtherGU/DUNormal

Handwritten

Reportmethod

Dictated

Computer

Error Bars shows95,0% CL of Mean

Bars shows Means

Scopist involved

time in seconds

Diagnosis

Endoscopistinvolvedtimeforuppergastrointestinalendoscopiesfordifferentreportingmethods,groupedbydiagnosiswithoutselectingphotos.GU/DU=GastricandDuodenalUlcerdisease.

51

gaged with composing and finalizing a dictated report. This is comparable to com-puterized predefined reports without endoscopic image selection time (86 seconds).

According to the ESGE guidelines, a certain number of endoscopic images are needed to document the whole tract thoroughly.13 Printing on photo paper with ordinary photo printers, as done in handwritten and dictated reports, is expen-sive. For this reason endoscopists often don’t illustrate handwritten and dictated reports with the recommended number of images. For computerized predefined reports an unlimited number of endoscopic images can be stored in the database. This off course needs an investment in storage capacity and a large database could have impact on the performance of the program. The digital images are stored as ‘jpeg-file’ and comprise only about 100 kB per photo, making this a minor problem. These images can also be compared with images of previous examinations. Com-puter hardware and software for computerized predefined reports are expensive, but these costs can be saved on the costs of a typist.

In computerized predefined reports all descriptions are standardized text. The content of this text has been subject of discussion in a working group of gastroen-terologists in the Netherlands, which has reached consensus on this matter. The


Figure 2 Endoscopistinvolvedtimeforlowergastrointestinalendoscopies,groupedbydiagnosis

0

50

100

150

Diverticulosis

Diagnosis

Scopist involved

time in seconds

OtherNeoplasmNormal

Handwritten

Reportmethod

Dictated

Computer

Error Bars shows95,0% CL of Mean

Bars shows Means

Endoscopistinvolvedtimeforlowergastrointestinalendoscopiesfordifferentreportingmethods,groupedbydiagnosiswithoutselectingphotos.

52

reports are automatically linked to extended ICD-10 based code, the gastrointesti-nal endoscopic terminology coding (GET-C).12 These codes are stored in the database, which offers possibilities for automated statistic reports and research.

This study primarily reports on the time aspects of the three report writing methods. No conclusions can be made about the content of the reports, as this was not the subject of this study. It is however clear that codes associated with the computerized predefined reports allow automated statistical analysis of endoscopic findings, whereas the use of free text in handwritten and dictated reports does not.

This study demonstrates that composing a computerized report of endoscopic findings is not more time consuming than generating a dictated or handwritten report. The storage of all endoscopic findings, including endoscopic images in a database offers a potential advantage to the other described reporting systems. Be-sides this it is important to make the report available in the hospital information system or patient record at any time.

Chap ter 3

53

References



3 Logan JR, Klopfer KC. The use of a standardized terminology for comparison of free text and structured data entry. ProcAMIASymp. 2000:512-6.

4 Waye JD, Aabakken L, Alvarez S, Archila P, Bornman PC, Hashiba K, et al. Endoscopy re-ports, databases, and computers in 2001. GastrointestEndosc. 2001;53(7):838-9.

5 Crespi M, Delvaux M, Schapiro M, Venables C, Zwiebel F. Minimal standards for a comput-erized endoscopic database. Ad hoc Task Force of the Committee for Minimal Standards of Digestive Endoscopy of the European Society of Gastrointestinal Endoscopy (ESGE). AmJ-Gastroenterol. 1994;89(8 Suppl):S144-S53.

6 Shapiro M. Computerization of Endoscopic Reports. Endoscopy. 1992;24(Suppl 2):478-80.7 Bell DS, Greenes RA. Evaluation of UltraSTAR: performance of a collaborative structured

data entry system. ProcAnnuSympComputApplMed Care. 1994:216-22.8 Gouveia-Oliveira A, Raposo VD, Salgado NC, Almeida I, Nobre-Leitao C, de Melo FG. Lon-

gitudinal comparative study on the influence of computers on reporting of clinical data. Endoscopy. 1991;23(6):334-7.

9 Kuhn K, Gaus W, Wechsler JG, Janowitz P, Tudyka J, Kratzer W, et al. Structured report-ing of medical findings: evaluation of a system in gastroenterology. Methods InfMed. 1992;31(4):268-74.

10 Moorman PW, van Ginneken AM, Siersema PD, van der LJ, van Bemmel JH. Evaluation of reporting based on descriptional knowledge. JAmMed InformAssoc. 1995;2(6):365-73.

11 Groenen MJ, Kuipers EJ, Berge Henegouwen GP, Fockens P, Ouwendijk RJ. Computerisation of endoscopy reports using standard reports and text blocks. NethJMed. 2006;64(3):78-83.

12 Groenen MJ, Hirs W, Becker H, Kuipers EJ, van Berge Henegouwen GP, Fockens P, et al. Gastrointestinal Endoscopic Terminology Coding (GET-C): a WHO-approved extension of the ICD-10. DigestDisSci. 2006.

13 Rey JF, Lambert R. ESGE recommendations for quality control in gastrointestinal endos-copy: guidelines for image documentation in upper and lower GI endoscopy. Endoscopy. 2001;33(10):901-3.


55


2 Department of Gastroenterology and Hepatology, Erasmus MC, Rotterdam

3 School of Management, Rotterdam Polytechnic, Rotterdam

4 Rotterdam School of Management, Erasmus University, Rotterdam

5 Department of Gastroenterology and Hepatology, Albert Schweitzer Hospital, Dordrecht

6 Department of Internal Medicine, Ruwaard van Putten Hospital, Spijkenisse

A cost-benefit analysis of

endoscopy reporting methods:

handwritten,

dictated and computerized

End o s cop y. 2 0 0 9Jul ; 41 ( 7):6 03 - 9 .


ShivaAjodhia1,3

J.Y.F.(Finn)Wynstra4

WilcoLesterhuis5

Ernst-JanJ.H.MvandeWeijgert6

ErnstJ.Kuipers2

RobJ.Th.Ouwendijk1

Chap ter 4

57

4Abstract

Background and study aims Gastrointestinal endoscopy investigations are fre-quently requested by gastroenterologists, general practitioners and other physi-cians. In addition to the classic methods of report writing, several electronic endo-scopic report systems are currently available. The aim of the study was to evaluate the costs of three different ways of producing reports; by hand, by dictation, or by computer.

Methods Three methods of report writing were compared, with special attention to costs. The endoscopy process was analyzed, from arrival of the patient to sending the report to the referring doctor, and including production of endoscopic images or video, logging of used endoscopes and their disinfection, and storage costs for endoscopy data.

Results During the first 5 years, the mean costs per procedure were € 4.78 for hand-written, € 6.39 for dictated and € 8.90 for computerized reports. Due to deprecia-tion, after this initial period, the respective costs declined to € 4.37, € 5.20 and € 5.13, respectively. Despite high initial costs, a cost–benefit analysis already revealed a financial benefit from a computerized system after 3 years.

Conclusions The electronic production of an endoscopic report turned out to be the most expensive way of report writing during the first 5 years, due to high initial costs. After 5 years the costs of the different systems were comparable with each other. Cost–benefit analysis showed a positive financial benefit for computerized reports after 3 years.

58

Chap ter 4

Introduction

As gastrointestinal endoscopy is a visual examination, it is of utmost importance that the endoscopist’s findings should be reported to the referring doctor. Tradi-tionally this has been done by means of a handwritten report or a transcription of a dictated report. Nowadays, an increasing number of electronic report systems are becoming available.1 These provide easily readable, well-structured, and complete reports that fulfill all the recommendations of several gastrointestinal endoscopy societies.2-3 As an additional benefit, an electronic report system often makes the report available within the hospital information system (HIS) and thereby continu-ously accessible for doctors.

Computerized systems need high initial investment, for hardware, software, and linkage to other medical computerized systems. Till now, the potential finan-cial benefits of specific computerized systems have not been documented and evaluated, hindering effective consideration of implementation of such a system.

At present about 50% of hospitals in the Netherlands use a computerized en-doscopy report system, mostly the Endobase III developed by Olympus. In most hospitals without computerized endoscopy reporting, the reports are produced by dictation and transcription or are still written out by hand. To the best of our knowledge, so far there has been no cost–benefit analysis of the different methods of generating endoscopy reports.

In the current study, the overall costs and the benefits of these three methods of report writing have been documented and compared. In our comparison we hypothesized that computerized report writing, apart from its nonfinancial bene-fits, is cost-effective after high initial investment.

Methods

In this study we compared the overall costs of three different methods of gener-ating reports: written out by hand, by the clinician; dictation and transcription; and production using a computerized system. The study was performed in three different endoscopy units in the Netherlands, one for each method. These centers were comparable in several important respects, such as in the open-access char-acter of the endoscopy unit, in endoscopy equipment, and in patient population. In the hospital with handwritten reports, three endoscopists between them per-formed a total of 4600 endoscopies in two endoscopy rooms. In the hospital with dictated records, four endoscopists between them performed a total of 5400 pro-cedures in three endoscopy rooms, and in the hospital with computerized report-ing four endoscopists between them performed a total of 4950 procedures in three endo scopy rooms.

59

A cost benefit analysis of endoscopy reporting methods

The steps in the reporting process were investigated. These steps began with the preparation for the endoscopy session and, if necessary, retrieval of the reports of previous endoscopies, followed by selection of patient information and noting which endoscope was used. The next steps involved the endoscopist’s capturing and printing the endoscopic images and then generating the endoscopic report. The logging of the endoscope and its cleaning process was also documented. The last step was dispatch of the endoscopy report to the referring doctor.

All resource costs for this process, such as for personnel, workspace and patient record storage, computerization, software, license, and material, were document-ed and subsequently evaluated for the three methods.

The initial and recurring (annual) costs were distinguished, and total costs per year and per endoscopic examination were calculated. For the costs per exami-nation, a 5-year depreciation period was applied for the initial investment costs. Personnel costs were calculated on the basis of the time the persons involved were actually working on the report, multiplied by their costs per hour.

In order to compare the three reporting methods, time and personnel require-ments were described for each stage in the procedure, as listed in Tables 1-3.

Table 1 Steps,personnel,andtimeneededforhandwrittenendoscopyreporting.

Step in report process Performed by Time, min:sec

Preparation

1 Carboncopyreferralpaperstickered2 Papermedicalrecordwithreferralpapertoendoscopyunit

SecretaryNurse

0:050:05

Endoscopyprocedureandreportproduction

3*Reportcomposedoncarboncopypaper Declarationformcompleted4 Imagesprintedonphotopaper5 Allformsplacedinpapermedicalrecordorhandedovertopatientforgeneral

practitioner Newappointmentchecked6 Papermedicalrecordstored7 Carboncopyofreportplacedin“endoscopy”file

Endoscopist

EndoscopistSecretary

SecretarySecretary

1:58

0:140:43

0:150:15

Reprocessingofendoscope

a Patientdetailslogged,alongwithdate,timeendoscopenumber,endoscopistandnurses

b Atcentralsterilizationdepartment,loggingofendoscopeidentification,numberofdisinfectionmachine,nameofemployee,anddisinfectionprogramused

Nurse 0:15

0:15

Totaltime 4:05

* Dateofexamination,endoscopistname,descriptionoffindings,conclusion,advice,medication

60

Handwritten reports [Table1]The reason for endoscopy was filled in on a blank carbon copy form at referral. The patient’s records were retrieved from the archives at 1 or 3 days before the endoscopy procedure, depending on whether the archive was located inside or outside the hos-pital.

Endoscopic images could be captured on a color video printer (Sony UP-21MD). Not all endoscopies had image records.

Because paper patient records were needed in this setting, the costs of archiving patient records, consisting of room and filing cabinets in the outpatient clinic, hos-pital archive, and external archive, were also taken into account.

Dictated reports [Table2]Reports of previous endoscopic investigations could be obtained from the HIS, but without endoscopic images. The endoscopist dictated the report following the en-doscopic examination. For patients referred by a general practitioner the tape was transcribed by the typist immediately after the examination. Otherwise the report was typed after the endoscopy clinic. Typing of endoscopic reports took half of the working time of one typist (20 hours a week). Endoscopic images could be captured

Chap ter 4

Table 2 Steps,personnel,andtimeneededfordictatedandtranscribedendoscopyreport.

Step in report process Performed by Time, min:sec

Preparation

1 Appointmentcheckedandindicationformreceived2 Papermedicalrecordwithstickersuppliedtoendoscopyunit

SecretarySecretary

0:121:11


3 Dictationofreporttotapeafterendoscopy4 SelectionofpatientrecordfromHIS*beforetyping5 Reporttyped6 Reportsavedandprinted7 Reportreadandagreedbyendoscopist8 Reporthandedtopatientorputintointernalpost

EndoscopistSecretarySecretarySecretaryEndoscopistSecretary

1:040:392:510:050:220:05

Reprocessingofendoscope

a Stickerwithpatientdetailsgiventoendoscopedisinfectionoperativetoplaceinlogbook

b Afterprocedure,operativelogstheirownname,endoscopeidentification,andnumberofdisinfectionmachine

Nurse 0:05

0:30

Totaltime 7:04

* HIS,hospitalinformationsystem

61

on a color video printer (Sony UP-5200). Not all endoscopies were recorded with images.

Endoscopes were logged in the disinfection room. There was no logging of the disinfection process in this hospital.

Computerized reports [Table3]A computerized endoscopy report system was used in which reports of all previous endoscopies, including the endoscopic images, were retrievable from the system before the investigation. The relevant patient data were automatically transferred from the scheduling program of the HIS into Endobase III. On arrival, the patient’s information was brought up on the computer terminal. During the examination an unrestricted number of digital endoscopic images as well as video tracks could be captured. Endoscopic images or video tracks were stored in about 90% of all examinations.

The report could be composed in three ways: by using standard reports, by select-ing text blocks or by structured data entry.4 All of these automatically generated a report using an extended code system based on the International Classification of Diseases (ICD)-10, known as gastrointestinal endoscopic terminology coding (GET-C).5

Table 3 Steps,personnel,andtimeneededforcomputerizedendoscopyreporting.

Step in report process Performed by Time: min:sec

Preparation

1 Appointmentchecked,indicationformreceived,patiententeredintosystem2*FieldscompletedinEndobasebynurse

SecretaryNurse

0:110:20


3 Imagesselected,reportcomposedandprinted4 Reporthandedovertopatientorsentbyinternalpost

EndoscopistSecretary

1:490:05

C Reprocessingofendoscopes

a Endoscopeautomaticallylinkedtopatient,endoscopist,endoscopynurse,anddisinfectionmachine,inEndobase

b Disinfectionoperativeisrecognizedviabatchwithmicrochipbydisinfectionmachine

0:00

0:03

Totaltime 2:28

* Nameofendoscopynurse,referringdoctor,indication,andpremedication


62

The Endobase program was continuously linked to the disinfection machines by a network connection. All information concerning the disinfection process, the dis-infection machine and the employees involved was recorded. Thus, the entire his-tory of an endoscope from its disinfection to its use and disinfection again, could be quickly ascertained.

Results

The mean time for the entire process of report production and endoscope documen-tation was 7 minutes 4 seconds using dictation and transcription, 4 minutes 05 sec-onds for handwritten reports, and 2 minutes 28 seconds for computerized reporting with Endobase III.

Initial costsInitial costs are listed in Table 4. For handwritten reports, the initial costs in-cluded: (i) storage area including shelves to store paper patient records; (ii) material costs, including a color photo printer for image reproduction and two paper books for the documentation of endoscopes and their disinfection; and (iii) a computer system including software with a connection to the HIS, for finding and storing patient information and printing stickers.

Dictated reports had higher initial costs for computerization, with workstations for the typists, software and license for the SAP server. Other initial (material) costs were for dictaphones, tape players, tapes, and a color photo printer for endoscopic images.

Chap ter 4

Table 4 InitialcostsinEurosfordifferentmethodsofendoscopyreporting:handwritten,dictatedandtranscribed,andcomputerized.

Initial costs, €

Handwritten Dictation Computerized

Storageofrecords*Materialcosts†Computerization‡

00 42005 50004 325

07 35322 388 94 214

Totalinitialcosts 10 245 29 741 94 214

* Storageareaandequipmentforpapermedicalrecords.† Materialcostsareforphotoprintersandphotoprinterpaper,carboncopypaper,dictaphones,tapesand

tape-players.‡ Computerizationinitialcostsareforworkstations,servers,printers,etc.andforprimarysoftwarelicenses.

63

Computerized reporting had the highest initial costs because of the investment in computer hardware and software. Additionally, several links between the Endo-base system and the HIS had to be established. In total the creation of four Health Level Seven (HL-7) links was needed: (i) from the HIS to Endobase, to schedule the endoscopy sessions; (ii) from the HIS to Endobase, to obtain patient information; (iii) from Endobase to the HIS, for report storage; and (iv) from Endobase to the HIS, to add endoscopic images to the report.

Recurring costsThe highest costs were for personnel, and were similar for all three reporting meth-ods, with an annual cost, based on 5000 endoscopy reports, of € 13 233 for handwrit-ten, € 14 380 for dictated and € 10 709 for computerized reports [Table 5]. Handwrit-ten reports incurred personnel costs for the endoscopist who composed the report and for administrative staff who retrieved and stored medical records and who recorded data needed for finance purposes. Dictated reports incurred the person-nel costs for the endoscopist and the typist. The personnel costs for computerized reports were mainly for the endoscopist producing the report. It took in general less

Table 5 Annual(recurring)costsineurosforreportingon5000endoscopies,usingdifferentreportingmethods:handwritten,dictatedandtranscribed,andcomputerized.

Annual costs for 5000 endoscopies, €

Handwritten Dictation Computerized

Personnel

EndoscopistEndoscopynurseEndoscopecleanerSecretary

(10 663 389 197 1 984)

(7 215 50 700 6 415)

(9 179 1 530)

Totalpersonnelcosts 13 233 14 380 10 709

Workspaceandstorage 2 623 9 600*

Computerization 900† 7 355‡

Softwarelicenses 7 211

Materialcosts§ 5 983 1 110 361

Totalannualcosts 21 839 25 990 25 636

* Additionalworkspaceneededforsecretary† Back-upforsystem‡ Back-upandcostsforlinksbetweenEndobaseandhospitalinformationsystem§ Carboncopyforms(handwritten),photoprintingpaper(handwrittenanddictated),tapesfordictaphones

(dictated),printingpaperforreports,andpostagecosts


64

than 2 minutes to compose a computerized report and print it, depending on the findings during the examination; this was comparable to findings in a larger study where time aspects of report writing were studied.6

Besides personnel, the recurring costs for handwritten reports were largely for stor-age of endoscopic reports in paper patient records. The remaining cost was for ma-terials such as carbon copy forms and photo paper.

For dictated report writing, recurring costs were for the typists’ workspace and some information technology (IT) costs for back-ups. The other costs were for materials including dictaphone tapes, photo print paper, and normal print paper for reports.

Computerized report writing incurred recurring costs related to maintaining HL-7 links between the Endobase and the HIS; making back-ups; Endobase soft-ware licenses and service contracts; and print paper for reports. After 5 years of de-preciation, further investment would have been needed for computer hardware. This item was booked for € 5000.

Costs per reportTo calculate the cost per report, the yearly costs and the depreciation costs were added together. Given the 5 years’ depreciation period, reporting costs were calcu-lated for the first 5 years as well as after 5 years. Based on 5000 endoscopy proce-dures per year, the costs per report were € 4.78, € 6.39 and € 8.90 for handwritten, dictated, and computerized reports, respectively, during the first 5 years and € 4.37, € 5.20 and € 5.13 after 5 years [Figure 1].

Chap ter 4

Figure 1 Costsperreportduringandafterthedepreciationperiod,with5000and10 000endoscopyproceduresperyear.

0

1

2

3

4

5

6

7

8

9 Handwritten 5000

Handwritten 10000

Dictated 5000

Dictated 10000

Computerrized 5000

Computerrized 10000

year 8year 7year 6year 5year 4

Years

year 3year 2year 1

Cost perreport

65

BenefitsWe have now considered the costs of the three methods, but not yet the associated financial benefits that might arise for the hospital, including those beyond the en-doscopy process itself. Comparing the computerized method with dictation, typing costs can be saved.

For each of the first 5 years, the computerized system would have been € 12 541 more expensive (€ 44 479 – € 31 938), due to the substantially higher depreciation costs. After 5 years, all the initial investments have been fully depreciated and the small computer reinvestment in year 6 is taken as an incidental cost. Hence, from year 6 on, the difference in cost between dictated and computerized reporting is changing to the advantage of the latter. Moreover, there are some differential ‘in-come’ benefits that make the investment in computerized reporting attractive at an even earlier point.

In the hospital with dictated reporting, typically one secretary is assigned for half of a working week (0.5 full-time equivalent [FTE]) to transcribing endoscopic reports. While we have chosen to only attribute labor costs for the effective time the secretary spends on transcribing dictated reports (€ 6 415), the actual salary costs for a secretary working 0.5 FTE were € 14 134. In other words, a secretary effectively spends about 45% of the time on typing reports. In the transition to computerized reporting, the costs of the secretary’s nonproductive time were also considered to represent potential savings (€ 7 719). Hence, these additional savings have been in-cluded in Table 6.

Computerizing the endoscopic report process can improve the capturing of in-office procedures that are carried out but not documented. Based on other stud-ies, a 2% improvement in billing capture was estimated.7 For 5000 endoscopies, this represents a yearly increase of billing by 100 procedures; with a mean of € 87 per investigation, this would mean € 8700 additional annual revenue.

The difference in material costs of dictated compared with computerized report-ing, amounted to € 750.

Besides these benefits there were many other nonmaterial benefits that are hard to express in concrete financial terms (see Discussion).

Comparing the costs and the material benefits of changing from dictated to computerized report writing, a net annual benefit of the computerized system is already apparent after 1 year. In the fourth year this leads to a cumulative financial benefit for the computerized system. After 5 years, when all the initial costs are depreciated, the cumulative benefits rise to about € 20 000 yearly in an endoscopy unit with 5000 investigations yearly [Table 6].


66

Discussion

This is the first known study comparing the costs of different methods of generat-ing endoscopic reports. As suspected, the initial costs of a computerized system are much higher than conventional methods of endoscopic report writing. This has its impact on the costs per report, especially in the initial period, when these costs have to be depreciated. After a period of 5 years, the costs per report are similar for the three different ways of report writing. This calculation does not take into ac-count savings on other costs, such as the financial benefits mentioned above, due to a switch from conventional to computerized report writing. If 10 000 investiga-tions were to be performed annually, the costs per report change dramatically. The main advantage is for the computerized reports, where the costs per report decline from € 5.13 to € 3.67 after 5 years [Figure 1].

Dictation of endoscopy reports involves more people, resulting in a higher risk of errors. Checking the typed reports would require a third manual stage.

Previous studies on electronic medical records have shown a positive cost–ben-efit balance.7 Computerized reporting saves material, workspace, and personnel costs compared with handwritten or dictated reporting. Over years the costs of personnel and workspace might rise unequally compared with the main costs of hardware and software that are associated with computerized systems, and this might lead to even higher benefits for computerized reporting. In this study only

Chap ter 4

Table 6 Cost–benefitanalysisforchangefromdictatedandtranscribedreportingtocomputerizedreporting,basedon5000endoscopyproceduresperyear.Allamountsareineuros(€).

Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7 Total

Costs

Excessannualcosts,computerizedvs.dictationNewcomputers

12 541 12 541 12 541 12 541 12 541 -354 5 000

-354

Totalcosts 12 541 12 541 12 541 12 541 12 541 4 646 -354 66 997

Benefits

SavinginpersonnelcostsBillingbenefits

7 719 0

7 719 8 700

7 719 8 700

7 719 8 700

7 719 8 700

7 719 8 700

7 719 8 700

Totalbenefits 7 719 16 419 16 419 16 419 16 419 16 419 16 419 106 233

Netannualbenefits -4 822 3 878 3 878 3 878 3 878 11 773 16 773

Cumulativebenefits -4 822 -944 2 934 6 812 10 690 22 463 39 236 39 236

67

one computerized system was compared with the other ways of report writing. Other computerized systems would probably show similar benefits, but this has to be studied further.

With computerized and dictation systems the endoscopy report is available in the electronic medical record, and accessible to doctors whenever needed. Com-pared with handwritten reporting, this results in a saving for computerized report-ing on the personnel costs for retrieval of paper patient records. In other studies these personnel savings from easier retrieval of patient data are estimated as one FTE per year.8 Annual savings were also associated with the storage costs for paper patient records, accounting for € 2623 with handwritten reporting.

The reduction of billing inaccuracy when using a computerized system that automatically generates a billing code when a report is produced can only be esti-mated. The 2% value for the increase in billing is based on findings in other stud-ies where an increase between 1.5% and 5% was generated.7, 9-10 The mean price for an endoscopy of € 87 is based on different endoscopic procedures (upper, lower, and therapeutic) with different reimbursement fees (range € 46 to € 104).

Benefits related to materials are especially related to the cost of photo printing paper. A computerized system does not require the expensive photo printing paper needed with handwritten and dictated reports. Only storage capacity is required, which is not a big problem with files of about 150 kB per image. Digital video cap-turing with a storage requirement of 40 MB per minute is a more expensive op-tion. Newer high-definition digital images also need greater capacity (600 kB), but still only a small fraction of that needed, for example, for radiologic investigations. Digital storage of endoscopic images and video makes it possible to review these images at any time in the future. According to European Society of Gastrointestinal Endoscopy (ESGE) guidelines, capture of eight specific images is recommended for every endoscopic investigation.11 American Society for Gastrointestinal Endoscopy (ASGE) quality guidelines require photographic documentation if an abnormality is encountered.12 If these guidelines are followed, images should be stored for at least 70% of investigations, this being the proportion in which any abnormality is found.13 With printed images this would lead to costs ranging from € 6125 (3500 in-vestigations [75% of 5000] with at least one sheet of photo paper, at € 1.75 per sheet), to € 12 250 (eight endoscopic images printed on two photo paper sheets, with up to 4 images per sheet).

Several benefits cannot be readily expressed in concrete financial terms. For example, the continuous availability of a report for a doctor in the HIS improves patient care. This makes debate about duration of storage of patient records redun-dant and is conducive to efficiency in the endoscopy department as well as through-out the hospital.

Other quality aspects concern the completeness of endoscopy reports. Studies


68

have shown that computerized structured systems produce superior reports com-pared with free text.14

The risk of transmission of infection between patients during an endoscopy pro-cedure is low.15 However, if there is suspicion of infection of an endoscope, the pa-tients in whom this endoscope had been used could be traced in less than a minute in the computerized system, in contrast to the other two systems, where most of the history was handwritten. Additionally, the risk of transmission is diminished by the use of an automatic monitoring system that makes it impossible to use an endoscope that has not been thoroughly cleaned and disinfected without a notifi-cation in the Endobase program.

This study has shown that computerized endoscopy reporting is cost-effective after an initial high investment, especially in hospitals where more than 5000 ex-aminations are performed annually. Besides cost-effectiveness, computerized re-porting shows many benefits that cannot be directly expressed in financial terms. All these advantages have to be taken into account when the purchase of a comput-erized endoscopy report system is considered.

Chap ter 4

69

References

1 Nelson DB, Block KP, Bosco JJ, Burdick JS, Curtis WD, Faigel DO, et al. Technology status evaluation report: computerized endoscopic medical record systems: November 1999. Gas-trointestEndosc. 2000;51(6):793-6.

2 Crespi M, Delvaux M, Schapiro M, Venables C, Zwiebel F. Minimal standards for a comput-erized endoscopic database. Ad hoc Task Force of the Committee for Minimal Standards of Digestive Endoscopy of the European Society of Gastrointestinal Endoscopy (ESGE). AmJ-Gastroenterol. 1994;89(8 Suppl):S144-S53.

3 Shapiro M. Computerization of Endoscopic Reports. Endoscopy. 1992;24(Suppl 2):478-80.4 Groenen MJ, Kuipers EJ, Berge Henegouwen GP, Fockens P, Ouwendijk RJ. Computerisa-

tion of endoscopy reports using standard reports and text blocks. NethJMed. 2006;64(3): 78-83.

5 Groenen MJM, Hirs W, Becker H, Kuipers EJ, van Berge Henegouwen GP, Fockens P, et al. Gastrointestinal Endoscopic Terminology Coding (GET-C): A WHO-approved extension of the ICD-10. DigDisSci. 2006:1004-8.

6 Soekhoe JK, Groenen MJ, van Ginneken AM, Khaliq G, Lesterhuis W, van Tilburg AJ, et al. Computerized endoscopic reporting is no more time-consuming than reporting with conventional methods. EurJInternMed. 2007;18(4):321-5.

7 Wang SJ, Middleton B, Prosser LA, Bardon CG, Spurr CD, Carchidi PJ, et al. A cost-benefit analysis of electronic medical records in primary care. AmJMed. 2003;114(5):397-403.

8 Neubauer AS, Priglinger S, Ehrt O. [Electronic versus paper-based patient records: a cost-benefit analysis]. Ophthalmologe. 2001;98(11):1083-8.

9 Pifer EA, Smith S, Keever GW. EMR to the rescue. An ambulatory care pilot project shows that data sharing equals cost shaving. HealthcInform. 2001;18(2):111-4.

10 Renner K. Cost-justifying electronic medical records. HealthcFinancManage. 1996;50(10): 63-4, 6, 8.

11 Rey JF, Lambert R. ESGE recommendations for quality control in gastrointestinal endos-copy: guidelines for image documentation in upper and lower GI endoscopy. Endoscopy. 2001;33(10):901-3.

12 Cohen J, Safdi MA, Deal SE, Baron TH, Chak A, Hoffman B, et al. Quality indicators for esophagogastroduodenoscopy. GastrointestEndosc. 2006;63(4 Suppl):S10-S5.

13 Groenen MJM, Ouwendijk RJT, Middelkoop MPC, Dees A, Bac DJ, van Berge Henegou-wen GP. The yield of open access upper gastrointestinal endoscopy. Evaluating 5943 upper endoscopies using Endobase© a standardised reportsystem European Journal of Gastro-enterology & Hepatology. 1999;11(12):A92.

14 Gouveia-Oliveira A, Raposo VD, Salgado NC, Almeida I, Nobre-Leitao C, de Melo FG. Lon-gitudinal comparative study on the influence of computers on reporting of clinical data. Endoscopy. 1991;23(6):334-7.

15 Nelson DB. Infectious disease complications of GI endoscopy: part II, exogenous infec-tions. GastrointestEndosc. 2003;57(6):695-711.


71



3 WHO-FIC Collaborating Centre in the Netherlands, RIVM/cVTV, Bilthoven

4 Department of Family Practice, Academic Medical Center, Amsterdam

5 Department of Gastroenterology, University Medical Center Utrecht, Utrecht

6 Department of Gastroenterology, Academic Medical Center, Amsterdam

Gastrointestinal Endoscopic

Terminology Coding

(GET-C): a WHO-approved extension

of the ICD-10

DigDisS ci. 2 0 0 7A p r ;5 2 (4): 1 0 0 4 - 8 .


WillemHirs3

HenkBecker4

ErnstJ.Kuipers1


PaulFockens6

RobJ.Th.Ouwendijk2

Chap ter 5

73

Abstract

Purpose Technological developments have greatly promoted the interest in the use of computer systems for recording findings and images at endoscopy and create databases. The aim of this study was to develop a comprehensive WHO-approved code system for gastrointestinal endoscopic terminology.

The International Classification of Diseases 10th edition (ICD-10) and the ICD-10 clinical modification (ICD-10-CM) were expanded to allow description of every pos-sible gastrointestinal endoscopic term under conditions defined by the WHO.

Results Classifications of specific gastrointestinal disorders and endoscopic loca-tions were added. A new chapter was developed for frequently used terminology that could not be classified in the existing ICD-10, such as descriptions of therapeu-tic procedures. The new extended code-system was named Gastrointestinal Endo-scopic Terminology Coding (GET-C).

Conclusions Gastrointestinal Endoscopic Terminology Coding (GET-C) is a com-plete ICD-10 related code system that can be used within every endoscopic database-program for all specific endoscopic terms. The GET-C is accessible for free on http://www.trans-it.org/.

5

74

Chap ter 5

Introduction

Technological developments and the introduction of hospital information systems have strongly promoted the interest in the use of computer systems for recording findings and images at endoscopy. Several database systems have been developed for this purpose, most only working as report generator. These systems are how-ever hampered by the lack of a specific code system for endoscopic diagnoses and terminology. As a result of this shortcoming, the composition of reports and stor-age of data differ considerably between systems. The systems use different database structures and variable terminology and are thus not compatible or comparable.

The International Classification of Diseases (ICD) is designed to promote inter-national comparability in the collection, processing, classification, and presen-tation of mortality statistics. This ICD is developed, published and maintained by the World Health Organization (WHO). The system is, in its current setting, not applicable for endoscopy as it does not include descriptive terminology. However endoscopic findings are classified according to their appearance. The adequate description of this appearance in uniform terminology is essential for the inter-pretation of endoscopy reports, and allows deriving important conclusions with respect to therapy and prognosis. This makes it important to uniformly store and code endoscopic findings.

This is however hampered by the lack of a specific, uniformly applicable code system for endoscopic terminology and findings. Besides that, only about 80% of the report systems are able to use a code in their database.1

Different code systems for medical data are being used. The International Clas-sification of Diseases 9th version (ICD-9) and the clinical modification (ICD-9CM) have already been used and proven their value in a specific and defined gastrointes-tinal endoscopic setting.2

The ICD has been revised periodically to incorporate changes in the medical field. The tenth revision (ICD-10) differs from the ninth revision (ICD-9) in several ways although the overall content is similar. First, ICD-10 has alphanumeric cat-egories rather than numeric categories. Second, some chapters have been rear-ranged, some titles have changed, and conditions have been regrouped. Third, ICD-10 has almost twice as many categories as ICD-9. The new version, International Classification of Diseases 10th version (ICD-10), has been available since 1992 and was translated into Dutch in 1997.3 Other medical specialties, such as oncology and den-tistry have produced an extended code system based on the ICD-10 for their field of interest.4

The aim of this project is to develop a comprehensive code system for gastrointes-tinal endoscopic terminology, based on the widely accepted ICD-10 code system, to be used in any gastrointestinal endoscopic information system.

75

Gastrointestinal Endoscopic Terminology Coding (GET-C)

Methods

As basis for the new code system the latest ICD-10th version was chosen. The reason for choosing this code system was that a predecessor has already been in use for some gastrointestinal endoscopic databases.2 Besides that, the new version of ICD has now been accepted in a majority of the countries throughout the world.5

Extensions to the ICD-10 are made according to the recommendations of the WHO, without changing the structure of the ICD-10. The first 4 characters are never changed, but one or more characters are added behind those first four descriptors. In this way it is always possible to delete the extension to come back to the original ICD-10. We used the Dutch translation of the ICD-10. Also the draft version of the ICD-10 CM of the National Center for Health Statistics (NCHS) was studied, which is available on the Internet.6 Conflicts with this ICD-10 CM system were avoided, to rule out coding conflicts.

A working group, the Trans.it-working group, was founded comprising of rep-resentatives from two academic hospitals, (Utrecht, Amsterdam) and 4 general hospitals. This working group used the Endobase III® system (Olympus). In this system, the endoscopist can choose between three different ways of report writing. Besides the Minimal Standard Terminology (MST) 7 also Text-blocks and Standard Reports are available. A standard report is a complete report based on a diagnosis or a combination of diagnoses. With text-blocks, different parts of text are combined to compose a complete report. These two variants enable a more rapid generation of reports, and are well accepted and widely used in the Netherlands.8

In the Trans.it-project group, which was founded as a peer reference group, the standard reports, text-blocks and the link to the new code-system were discussed.

A list of every endoscopic finding, intervention or complication described with-in one of the three different ways of report writing available for Trans.it users was generated. All the items on this list were linked to a specific code.

Results

We first generated a list of different fields necessary for evaluation of endoscopic data [Table 1]. The program automatically generated the data during composition of endoscopic reports. All these fields are necessary to generate a comprehensive anonymous database with endoscopic information for extensive research. In addi-tion, demographic data like gender and age of patients, specialty of referring doc-tor, etc are recorded.

76

We included all available standard reports, text-blocks and diagnoses of MST ac-cepted by the Trans.it-group. The endoscopic findings, interventions and compli-cations were extracted and linked to a specific code. In this way, a total of 316 stan-dard reports and 1571 text-blocks were coded, to assure coding of every item within this project.

In total 2593 different items were extracted and received a specific code. Of these items, 630 (24%) could be coded with the existing ICD-10. Thus, 1963 new codes were required and added to the ICD-10 system, in order to be able to specifically code every endoscopic term.

The used ICD-10 was originally designed for mortality and discharge statistics. The problem of that design is that the used structure in the ICD-10 is not very de-tailed. This is especially for gastrointestinal endoscopic terms, which do not always contain a mortality risk but most of the time descriptive items essential for progno-sis and therapy. For example, specific descriptions of peptic ulcers are relevant for determination of outcome risks, therapy and prognosis. We therefore introduced the Forrest Classification for description of gastric and duodenal ulcers.9 We thus for instance changed the diagnosis K25.2, (Acute gastric ulcer with both haemor-rhage and perforation) into K25.21 (Acute gastric ulcer with spurting bleeding and per-foration; Forrest Ia) in the GET-C [Table 2].

Besides such an adaptation of the ICD-10 system, endoscopy is still developing and new terms and endoscopic techniques are continuously being created. For exam-ple, new disease classifications are from time to time being introduced, such as the semi recent LA-Classification for reflux esophagitis.10

For several fields, there was no ICD-10 code available, and the item could thus not be categorized under an existing ICD-10 code. For example, the proceeding of the investigation and interventions could not be covered within the ICD-10 and even the

Chap ter 5

Table 1 Differentfieldsthatneedaspecificcode

FieldsinendoscopicinformationsystemsthatneedaspecificcodeReasonforendoscopyMedicationuseSedationandmedicationduringtheendoscopyPreparationProceedingoftheinvestigationEndoscopicdiagnosis/findingsTherapeuticandDiagnosticInterventionsHistologyresultsTherapystartedAdvicetoreferringdoctorComplications

77

ICD-10 CM system. Because these items are essential for good analysis of endoscopic data, a new chapter with the same structure of the ICD-10 was developed. The items in this new chapter start with the Greek letter μ. We chose to categorize the thera-peutic interventions in this new chapter also because new therapies are developed frequently so they can be adapted.

Adaptations were made in different chapters of the ICD-10.


Table 2 ICD-10andGET-CforGastricUlcer

ICD-10 Description GET-C Description

K25.0 Gastriculcer,acutewithhaemorrhage

K25.0K25.01K25.02

Gastriculcer,acutewithhaemorrhageGastriculcer,acutewithspurtingbleeding(ForrestIa)Gastriculcer,acutewithnon-spurtingactivebleeding(ForrestIb)

K25.1 Gastriculcer,acutewithperforation

K25.1 Gastriculcer,acutewithperforation

K25.2 Gastriculcer,acutewithbothhaemorrhageandperforation

K25.2K25.21K25.22

Gastriculcer,acutewithbothhaemorrhageandperforationGastriculcer,acutewithspurtingbleedingandperforation(ForrestIa)Gastriculcer,acutewithnon-spurtingactivebleedingandperforation(ForrestIb)

K25.3

Gastriculcer,acutewithouthaemorrhageorperforation

K25.3K25.31K25.32K25.33K25.34

Gastriculcer,acutewithouthaemorrhageorperforationGastriculcer,acutewithvisiblevessel(ForrestIIa)Gastriculcer,acutenon-bleedingwithoverlyingclot(ForrestIIb)Gastriculcer,acutewithhematin-coveredbasis(ForrestIIc)Gastriculcer,acutewithcleanulcerground(ForrestIII)

K25.5 Gastriculcer,chronicorunspecifiedwithperforation

K25.5 Gastriculcer,chronicorunspecifiedwithperforation

K25.6

Gastriculcer,chronicorunspecifiedwithbothhaemorrhageandperforation

K25.6K25.61K25.62

Gastriculcer,chronicorunspecifiedwithbothhaemorrhageandperforationGastriculcer,chronicorunspecifiedwithspurtingbleedingandperforation(ForrestIa)Gastriculcer,chronicorunspecifiedwithnon-spurtingactivebleedingandperforation(ForrestIb)

K25.7 Gastriculcer,chronicwithouthaemorrhageorperforation

K25.7K25.71K25.72K25.73K25.74

Gastriculcer,chronicwithouthaemorrhageorperforationGastriculcer,chronicwithvisiblevessel(ForrestIIa)Gastriculcer,chronicnon-bleedingwithoverlyingclot(ForrestIIb)Gastriculcer,chronicwithhematin-coveredbasis(ForrestIIc)Gastriculcer,chronicwithcleanulcerground(ForrestIII)

K25.9

Gastriculcer,unspecified,withouthaemorrhageorperforation

K25.9K25.91K25.92K25.93K25.94

Gastriculcer,unspecified,withouthaemorrhageorperforationGastriculcer,unspecified,withvisiblevessel(ForrestIIa)Gastriculcer,unspecified,non-bleedingwithoverlyingclot(ForrestIIb)Gastriculcer,unspecified,withhematin-coveredbasis(ForrestIIc)Gastriculcer,unspecified,withcleanulcerground(ForrestIII)

78

In Chapter 1, Certain infectious and parasitic diseases (A00-B99), the exact loca-tions in the gastrointestinal tract of some specific infections were added.

In Chapter 2, Neoplasm’s (C00-D48), besides exact locations in the gastrointesti-nal tract the endoscopic characteristics of polyps were described and coded in more detail. It is important to register if a polyp is pedunculated, sessile, flat or has a villous endoscopic aspect. Also the number of polyps found in a specific part of the intestine has to be coded, because of important diagnostic and therapeutic options for the patient and the follow-up of these patients.

In Chapter 9, Diseases of the circulatory system, (I00-I99), hemorrhoids and var-ices were described according to a classification. Different gradings of protrusion were used for Haemorrhoids. The endoscopic Paquet’s classification for esophageal varices was coded.11

Most changes were however made in chapter 11, Diseases of the digestive sys-tem (K00-K93). The description of esophagitis was divided in different origins and severity. Gastric and duodenal ulcers were coded according to the Forrest classifica-tion and their specific location in stomach or duodenum. Gastritis and duodenitis were coded to their specific location and the endoscopic suspected cause. Inflam-matory bowel diseases were coded according to their endoscopic severity and loca-tion. Some complications related to gastrointestinal procedures were more coded in more detail.

In Chapter 18, Symptoms, signs and abnormal clinical and laboratory findings that are an indication for endoscopic investigations were coded in more detail.

In Chapter 21 post-operative situations important for gastrointestinal endoscop-ic investigations like gastric and colon operations were extensively coded.

In the new 13th chapter of the GET-C, about 446 new codes were generated. These codes start with the Greek symbol μ. The first part of this chapter including coding of various indications for endoscopy that could not be categorized in the ICD-10. The second part consisted of specific codes for detailed locations in the gastrointestinal tract. The third part of this chapter contained procedures, which were divided in diagnostic and therapeutic procedures. Diagnostic procedures contained for exam-ple sampling of histology and culture specimens.

Examples of therapeutic procedures were different kinds of polypectomies, endo scopic mucosal resections, the placement of endoprotheses, and dilatation of stenoses. Another part of these therapeutic interventions was used for the manage-ment of gastrointestinal bleeding like injection-therapy with or without coagula-tion, band ligations, and clipping.

Preparation and proceeding of the endoscopic examination can also be coded in this chapter.

The extensions of the GET-C were checked by the Dutch translation board of the ICD-10 to preclude any conflicts between the two code systems.

Chap ter 5

79

Discussion

The increased use of computer systems within healthcare and the need for com-munication between these systems necessitate the availability of generally ac-cepted code systems. Good coding gives epidemiological information for research purposes. Coding is of importance for hospitals and professionals because most of the financial systems within healthcare are based on different codes. In addi-tion, statistical analysis of different diagnoses and the collection of rare diagnoses are easier. With international acceptation of a coding it is possible to compare and share information in one field of interest.

Most code systems are regional or national. Only some code systems, like the ICD-10, are translated into different languages and in use in different countries.

For endoscopic report writing and endoscopic databases, different systems have been developed. These systems are now being used in many hospitals. We used the Endobase III system from Olympus Europe in our project to compose reports and record different codes. The system runs as a network-version with different work-stations as well as a stand-alone unit.

The system enables the generation of reports via three different pathways; be-sides standard reports and text-blocks also the latest translated version of MST is used to compose an endoscopic report. This makes it essential that all different gen-erated data are coded in the same way to be able to analyze the data anonymously.

Within the system, all standard reports and text-blocks are linked to one or more specific GET-C codes. The choice of a standard report or text-block directly leads to the recording of the correct GET-C code in the Endobase database. For the MST, the automatic link is more complex to realize, due to the structure of the MST. We chose to link the diagnoses with the GET-C code, which is separately selected within the MST by the endoscopist at the end of the report. Automatically linking is essential to ensure a correct and complete selection of codes and to make it workable for an endoscopist.

Within the Trans.it project, an anonymously central database of endoscopic investigations was build. In this central database, only coded data instead of free text can be collected for privacy reasons. After having been used for three years, a database of about 120.000 investigations with the same code system will be build by this working group.

The GET-C system allows encoding any data collected during endoscopic inves-tigations and recorded in any endoscopic database. By extending the ICD-10 in respect to the structure developed by the WHO it is always possible to retract the original ICD-10 code out of the GET-C. Thus, it is possible to link the endoscopic database with other healthcare systems throughout the world.

Because endoscopy is an evolving medical specialty, new techniques are contin-


80

uously becoming available. This requires new codes. These codes will be discussed within the Trans.it project.

The Dutch Association of Gastroenterologists in the Netherlands has accepted the GET-C.

The GET-C will be available for use in other systems, and has been discussed at the WHO Family of international classifications network meeting in October 2004.12 The GET-C is accessible for free on http://www.trans-it.org. We hope that this code system will be helpful in making endoscopic databases and endoscopic report writ-ing programs more valuable.

Chap ter 5

81

References

1 Teuffel W, Stettin J. Electronic documentation in endoscopy: present status and future perspectives from a company standpoint. Endoscopy. 2001;33(3):276-9.

2 Cooper GS, Chak A, Lloyd LE, Yurchick PJ, Harper DL, Rosenthal GE. The accuracy of diag-nosis and procedural codes for patients with upper GI hemorrhage. GastrointestEndosc. 2000;51(4 Pt 1):423-6.

3 Who, Nrv WCC. Internationale Statistische Classificatie van Ziekten en met Gezondheid verband houdende Problemen Tiende Revisie: CSIZ; 1997.

4 WHO. International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3). Available from: http://www.who.int/classifications/icd/adaptations/oncology/en/index.html.

5 WHO. Implementation of ICD. Available from: http://www.who.int/classifications/icd/implementation/en/index.html.

6 Statistics CNCfH. Classification of Diseases, Functioning, and Disability. Available from: http://www.cdc.gov/nchs/icd.htm.

7 Delvaux M, Crespi M, Armengol-Miro JR, Hagenmuller F, Teuffel W, Spencer KB, et al. Minimal standard terminology for digestive endoscopy: results of prospective testing and validation in the GASTER project. Endoscopy. 2000;32(4):345-55.

8 Groenen MJ, Kuipers EJ, Berge Henegouwen GP, Fockens P, Ouwendijk RJ. Computerisa-tion of endoscopy reports using standard reports and text blocks. NethJMed. 2006;64(3): 78-83.


10 Lundell LR, Dent J, Bennett JR, Blum AL, Armstrong D, Galmiche JP, et al. Endoscopic as-sessment of oesophagitis: clinical and functional correlates and further validation of the Los Angeles classification. Gut. 1999;45(2):172-80.

11 Paquet KJ. [Diagnostic and therapeutic measures in acute catastrophic bleeding esopha-geal varices]. FortschrMed. 1976;94(33):1941-6.

12 WHO Family of International Classifications Network Meeting. 2004; Available from: http://www.nordclass.uu.se/WHOFIC


83


2 Department of Gastroenterology and Hepatology, Rijnstate Hospital, Arnhem

3 Department of Gastroenterology and Hepatology, University Medical Center Utrecht

4 Department of Gastroenterology and Hepatology, Academic Medical Center, Amsterdam

5 Department of Medical Informatics, Erasmus Medical Center, Rotterdam

6 Department of Gastroenterology and Hepatology, St. Elisabeth Hospital, Tilburg

7 Department of Gastroenterology and Hepatology, Atrium Medical Center Parkstad, Heerlen

8 Department of Gastroenterology and Hepatology, Ikazia Hospital, Rotterdam, The Netherlands

Validation study of

automatically generated codes

in colonoscopy using the endoscopic

report system Endobase®

S c a ndJ G a s t r oen t er ol . 2 0 1 0Sep ; 45 ( 9): 1 12 1 - 6


HenkR.vanBuuren1


PaulFockens4

JohanvanderLei5

WouterN.H.M.Stuifbergen6

PeterJ.vanderSchaar7

ErnstJ.Kuipers1

RobJ.Th.Ouwendijk8

Chap ter 6

85

Abstract

Objective Gastrointestinal endoscopy databases are important for surveillance, epidemiology, quality control and research. A good quality of automatically gener-ated databases to enable drawing justified conclusions based on the data is of key importance. The aim of this study is to validate the correctness of coding of a na-tional automatically generated anonymous endoscopy database.

Material and Methods We evaluated a total of 500 colonoscopies performed in five larger hospitals of the Trans.it project focusing on endoscopy reporting. Random-ly 500 examinations were selected from a total of 5,000 examinations and their generated endoscopic terminology codes as well as complete reports were analysed. Indications for the examination and described findings were scored for correctness and clinical relevance of the coding that would be exported to the anonymous da-tabase.

Results Indications were correctly coded in 92% of all examinations (range 76-100%) per hospital. Correct coding of findings ranged from 42% to 93% per hospital (mean 77%). Different correct coding proportions were seen varying with the diagnosis, with the highest correct coding rates in polyps, carcinoma and diverticular dis-ease. Incorrect coded examinations were scored for clinical relevance. Overall 11% of the investigated examinations were incorrectly coded with clinical relevance.

Conclusions Accuracy of clinical relevant endoscopy data recorded in the Trans.it anonymous central database is high. Further improvement is desirable, which may be achieved by education of individual endoscopists and enhancement of the program.

6

86

Chap ter 6

Introduction

Gastrointestinal endoscopy has become a routine, widely available technique for diagnosis and treatment of gastrointestinal disorders. A recent survey among en-doscopists in the Netherlands showed that more than 400,000 gastrointestinal en-doscopies were performed annually in a population of over 16 million.1 These num-bers can likely be extrapolated for the European Union, which means that the EU population receives more than 12 million endoscopic investigations per year.

An imaging method with numerous repetitive manoeuvres and findings, is par-ticularly suitable for electronic storage and processing of data.2 Computerized en-doscopic databases may play an important role in quality control and patient care, including monitoring of complications and outcome as well as identification of individuals eligible for endoscopic surveillance procedures. Endoscopic databases may also allow and facilitate epidemiological and other types of scientific research and may be very helpful in teaching.

For these purposes, a number of systems have been developed for recording endoscopic findings, storing images and composing reports.3-13 Unfortunately how-ever, most of these are stand-alone report systems which cannot be integrated into general hospital information systems. Other disadvantages of most systems include the inability of combined report writing and digital image storing, poor accessibility of stored data for research and limited or absent possibilities for ex-tracting or exporting data.

In close to 50% of all hospitals in the Netherlands, one particular endoscopy report and database system (Endobase®, Olympus, Software) is used.

This Endobase system offers different methods of report writing for composing an endoscopic report in a structured way. For all gastrointestinal endoscopic inves-tigations a Dutch version of this medical information was developed. Besides stan-dard reports, text blocks and structured data entry can be used for more complex or multiple diagnoses.14-15

Endobase® (nowadays named Endoalpha Documentation®, Olympus, Software) was built around an extensive relational database structure. In this way it becomes suitable to store all data produced in an endoscopy unit, including patient charac-teristics, patient monitoring data, digital images and video, endoscope informa-tion, and complications. These data can be separately retracted with all kinds of queries.

In 1999 the Trans.it group, a working group consisting of members from three university and seven general hospitals, was founded to develop a uniform, consen-sus-based endoscopy report writing system using Endobase®. A secondary aim was to establish a system of automated linkage of the composed report to an extended diagnostic coding system based on the tenth version of the International Classifi-

87

Validation study of automatic codes in colonoscopy

cation of Diseases (ICD-10). This Gastrointestinal Endoscopic Terminology Coding(GET-C) is automatically linked to the selected standard reports and text blocks entry when assembling a report.16 The indications for the examinations are impor-tant details for analysing outcome of endoscopy investigations and are also auto-matically coded by the system.

Besides report making, Endobase also facilitate direct financial and adminis-trative tasks related to the endoscopic procedure.17

Although the Endobase system for report making is based on the use of struc-tured data entry, text blocks or (nearly) complete reports, the reporter can add free text or adjust the automatically generated report. Obviously, this may possibly in-terfere with and impair the adequacy of the automated terminology coding. As the Endobase system is now becoming a standard for endoscopy reporting and thus also serves as an example for other systems, it is relevant to evaluate to assess the consistency of the data generated by such a system which is of key importance for future research. Therefore, the present multicenter study was performed to assess to what extent automatically generated GET-C codes are in agreement with the ac-tual text of the endoscopy report.

Methods

Type and selection of reportsThis was a retrospective study involving five hospitals (2 university and 3 general hospitals) employing the Endobase report system and participating in the Trans.it endoscopic database project. The endoscopists in the five hospitals had agreed to use both standard reports and text blocks.

For reasons of variability of findings the study was performed on the results of colonoscopies. The last 1000 colonoscopies, performed before January 2006 in every hospital, were retrieved from the endoscopic database. The name of the en-doscopist, indication, medication, selected standard reports and text blocks titles, GET-C codes, and the complete anonymous final report were exported. From these 1000 cases, a random sample of 100 procedures per hospital was obtained with the random sample function of Statistical Package for the Social Sciences version 14.0 (SPSS) from SPSS Inc. Chicago, Illinois.

DefinitionsThe final complete endoscopy report was considered to be the gold standard. An examination was considered appropriate if all the abnormalities in the final report were correctly coded in the database. Three experienced gastroenterologists, who were blinded for hospital and endoscopist data, scored every examination for

88

correctness of the GET-C codes and indications for the examinations considering the complete contents of the final report. If any of the three reviewers found any discrepancies between the automatically generated GET-C codes and the endoscopy report text, the procedure was scored as incorrectly coded. If an abnormality was described in the final report but not coded, it was classified as a false-negative cod-ing. If there was a code for a specific abnormality, but this was not described in the final report, it was classified as a false-positive coding.

In the same way, incorrectly coded examinations were classified as clinically relevant or irrelevant. The results were analyzed not only per hospital but also per examiner, to get a impression of inter-examiner variability.

Indications for the examination were scored as false-negative if an indication was not recorded in the database but was only written in the report and as false-positive if an indication has been selected but was not present in the final report.

Results

Hospitals and endoscopistsAll hospitals performed at least 1000 colonoscopies in a period of seven months or less. The randomly selected 100 examinations per hospital all showed valid infor-mation to allow complete analysis.

A total number of 45 endoscopists were involved; the number of endoscopists per hospital varied from 4 to 20. In 385/500 (77%) procedures the GET C codes corres-

Chap ter 6

Figure 1 Percentageofcorrectcodingin5differenthospitals

Hospital

PercentageCorrect Coding

0

20

40

60

80

100

51423

42

70

899193

89

ponded completely with the endoscopy report text. These percentages consistent reports varied between hospitals from 42% to 93% [Figure 1]. Differences in proportions of correct recordings between endoscopists ranged from 0% to 100% [Figure 2]. It should be noted that the number of procedures per endoscopists ranged markedly, with a minimum of one and maximum of 42. If only endoscopists, who performed more than 5 investigations in the present sample, were included the proportion of colonoscopies with concordant data had a mean of 74%.

Analysis showed that results of one particular hospital were markedly different from those of the other centres. In this single hospital, only 42% of examinations were correctly coded, as compared with 86% (range 70%-93%) for the other four hos-pitals. This can be explained by an incorrect way of report making. Although it was agreed to use standard reports and text blocks, in this hospital in the majority, reports were composed using free text.

Diagnosis groupsExclusion of the single, low-scores hospital revealed 400 remaining examinations for analysis. These were divided into six different diagnosis groups based on the most relevant findings: Normal, Polyps, Carcinoma, Diverticular disease, Inflam-matory Bowel Disease and Other [Table 1]. The percentage of procedures with cor-rect coding (81-91%) was not significantly different between the diagnostic catego-ries [Table 1].


Figure 2 Percentageofcorrectcodingperindividualexaminer(n=45)

Examiner

0

20

40

60

80

100

0

11

2513

4047

5057

6267

7175

8186878889

90949698100

PercentageCorrectCoding

90

The 57 incorrectly coded examinations (14% of total examinations) were scored as false-negative in 54 of the cases (95%) and false-positive in 3 cases. In these three cases extra information was coded in the database, but this was deleted from the final report by the endoscopist.

Evaluating the incorrectly coded examinations, for clinical relevance revealed 44 (77%) of the examinations to be clinically relevant. In the majority these were polyps, not corresponding number of polyps, diverticular disease, but also one rec-tal carcinoma was not coded.

Chap ter 6

Table 1 Mainendoscopicfindingsandpercentagecorrectcodingof4hospitals.(Hospital5excludedfromdataanalysis)

Main endoscopic finding Correct Coding

Number Percentage Percentage

Normal 176 44.0 81.3

Polyps 101 25.3 90.1

Carcinoma 19 4.8 89.5

DiverticularDisease 55 13.8 90.9

IBD 28 7.0 85.7

Miscellaneous 21 5.3 85.7

Total 400 100.0 85.8

Table 2 Numberandpercentageofincorrectcodingandtheirclinicalrelevance

Incorrect Coding Correct Coding

Number False negative Clinically relevant Percentage

Hospital1 30 28(93.3%) 19(63.3%) 70

Hospital2 9 8(88.9%) 7(77.8%) 91

Hospital3 7 7(100%) 7(100%) 93

Hospital4 11 11(100%) 11(100%) 89

Hospital5a

Total 57(14.2%) 54(13.5%) 44(11.0%) 85.8

a Excludedfromdataanalysis

91

Of the 57 incorrectly coded examinations 13 were not clinically relevant, containing incipience polyps in a report that was already coded for other polyps, some post-surgical situations, not further defined reddish mucosa, and grade-I haemorrhoids accompanied by already coded abnormalities. [Table 2]

IndicationsIndications were correctly recorded in the database in 92% of all examinations, ranging from 76.0 % to 100 % per hospital. All incorrectly coded indications were false-negative, due to not selecting an indication in the system and adding the in-dication in free text only.

Discussion

This study shows that automatically generated codes from gastrointestinal endo-scopy reports can firstly be very consistent between different endoscopy units, and secondly with high adequacy reflect the actual endoscopic findings.

A review has been performed on the accuracy of computer based patient records with a high variety in correctness and completeness of data,18 but little is known about the level of adequacy and inter-individual consistency of automatically gen-erated coding in endoscopy reporting. To our knowledge, there are no studies on reliability of automatically generated codes in endoscopic databases.

This study was performed in the Netherlands, with the objective to investigate the proportion of incorrectly recorded data in a large endoscopy report database, in order to show the usefulness of the database on the one hand and on the other hand to be able to compare these initial results with a planned intervention to improve the quality of database building in the future.

Medical databases are not only important for patient care, but can additionally be used for clinical research, health-system management, health-service plan-ning, total quality improvement, billing, risk management and government re-porting. The accuracy of these data is therefore of utmost importance.

The quality of the data in a database depends in the first place on the correctness of the input of data.

If data have to be entered manually, building of a substantial database requires a lot of time of the persons entering the data. However, using advanced comput-erized systems working with a structured data-input to compose reports enables database building in which, during daily busy practice, a lot of data are recorded and entered automatically. Most of these systems still offer the possibility to adapt text by free input. In this way the system is workable for almost every user. The op-


92

portunity to use free text enhances detailed description but at the same time creates the risk of incorrect structured data.19 Due to the possibility to remove information from the final report of already recorded data, false-positive data may be created. On the other hand, the possibility to describe abnormalities in free text without really recording it in the database, may result in false-negative data. The aim of the Trans.it working groups is to minimize the use of free-text in our endoscopy report system. This goal is probably not easily achieved nationwide, but shows to be pos-sible after instructions in participating high volume endoscopy centres.

Surprisingly, large differences are seen in correctness between hospitals and be-tween individual endoscopists. It was an eye-opener to see that one of the par-ticipating hospitals used another (and thus incorrect) way of reporting than was agreed upon in the working group. All endoscopists in this hospital appeared to record the majority of data incorrectly, suggesting a general problem in making the reports. They appeared to have used free text in the majority of their composed re-ports, making them unsuitable for generating codes. Considering the fundamen-tal difference with the other hospitals it was decided to exclude this hospital from further specific analysis.

When only the clinically relevant incorrect records were considered, only in 11% of the included examinations the recorded code was not consistent with the final report. This 89% correct coding is high, in comparison with other studies, but trig-gers a further investment to enhance the results for this specific database.18

For different diagnosis groups small differences were seen in correct coding rates. Particularly carcinoma and polyps, which are of extreme importance for di-agnosis as well as for follow-up, appeared to contain the best recorded data, with 90% correct data.

As expected, examinations coded as normal, showed the lowest percentage of correct codings. This is probably due to the use of free text to describe abnormalities in a standard report with normal findings.

Therefore adaptations that enable a quicker search for the corresponding text block and education for the endoscopist is planned in the future. Initiating feed-back information to the endoscopist of his incorrectly use of the report system will enhance quality of the database input. Additional training for endoscopists work-ing in these hospitals will be performed.

We think that these results are not only applicable for the Endobase system, but are likely to reflect other report systems that generate codes automatically. This hy-pothesis needs to be confirmed. Unfortunately no data exist to compare our results with other systems. Such a comparison, although seemingly tedious, is relevant as the use of these databases is rapidly increasing in daily patient care, hospital management, and research.

Chap ter 6

93

In conclusion, this study shows that overall a high proportion (89%) of clinically relevant data is recorded correctly, making this anonymous database suitable for analysis. However there are remarkable differences in correctness of generated codes between endoscopists. Therefore further improvement of the correctness of the generated data is considered necessary. We intend to induce better recording of endoscopy data in the future by showing each hospital and individual endoscopist their data, compared with the anonymous results of other hospitals. With new rec-ommendations, guidelines for composing the report and training of endoscopists we expect a higher proportion of data in the central database will be coded cor-rectly.


94

Chap ter 6

References

1 Terhaar Sive Droste JS, Craanen ME, Kolkman JJ, Mulder CJ. Dutch endoscopic capacity in the era of colorectal cancer screening. NethJMed. 2006;64(10):371-3.

2 Classen M, Wagner F, Swobodnik W. Electronic data base in gastroenterological endos-copy. Endoscopy. 1991;23(1):29-31.

3 Ariotti G, Allosia GP, Ballerini P, Corti E. Micro-ordinateur et fichier de protocoles d’endoscopie: coût et bénéfice d’un système de base de données. 1985;15(5): 419-29.

4 Cooper G, Sivak MV, Jr. More on endoscopy databases. GastrointestEndosc. 2000;51(6): 759-61.

5 Feczko PJ, Ackerman LV, Halpert RD, Simms SM. A computer-based gastrointestinal infor-mation management system. Radiology. 1984;152(2):297-300.


7 Kruss DM. The ASGE database: computers in the endoscopy unit. Endosc Rev. 1987;4:64-70.8 Reichel W. [Computer systems in gastroenterologic endoscopy. An analysis of current

availability on the Austrian market]. ZGastroenterol. 1993;31(3):183-91.9 Rozen P. Computer assistance in gastroenterology: an update. AmJGastroenterol.

1988;83(12):1323-7.10 Stoltzing H, Birkner B, Lindlar R, Ohmann C, Zaczyk R, Kuntzen O, et al. [Computer-

assisted documentation in upper gastrointestinal endoscopy: experiences with routine use at 3 clinics]. ZGastroenterol. 1989;27(11):667-75.

11 Waye JD, Aabakken L, Alvarez S, Archila P, Bornman PC, Hashiba K, et al. Endoscopy re-ports, databases, and computers in 2001. GastrointestEndosc. 2001;53(7):838-9.


13 Zimmerman J, Shapiro E, Porath Y, Rachmilewitz D. Computer application in a gastroin-testinal endoscopy unit. Endoscopy. 1985;17(6):207-9.


15 Soekhoe JK, Groenen MJ, van Ginneken AM, Khaliq G, Lesterhuis W, van Tilburg AJ, et al. Computerized endoscopic reporting is no more time-consuming than reporting with conventional methods. Eur J Intern Med. 2007 Jul;18(4):321-5.

16 Groenen MJ, Hirs W, Becker H, Kuipers EJ, Berge Henegouwen GP, Fockens P, et al. Gastro-intestinal endoscopic terminology coding (GET-C): a WHO-approved extension of the ICD-10. DigDisSci. 2007;52(4):1004-8.

17 Groenen MJ, Ajodhia S, Wynstra JY, Lesterhuis W, van de Weijgert EJ, Kuipers EJ, et al. A cost-benefit analysis of endoscopy reporting methods: handwritten, dictated and comput-erized. Endoscopy. 2009 Jul;41(7):603-9.

18 Hogan WR, Wagner MM. Accuracy of data in computer-based patient records. JAmMedIn-formAssoc. 1997;4(5):342-55.

19 Hogan WR, Wagner MM. Free-text fields change the meaning of coded data. Proc AMIA Annu Fall Symp. 1996:517-21.

95



3 Department of Epidemiology & Biostatistics, Erasmus Medical Center, Rotterdam

Incidence of duodenal ulcers

and gastric ulcers

in a western population;

back to where it started

C a nJ G a s t r oen t er ol . 2 0 0 9Sep ;23 ( 9):6 0 4 - 8


ErnstJ.Kuipers1

BettinaE.Hansen1,3

RobJ.Th.Ouwendijk2

Chap ter 7

97

Abstract

Background Over the past decades, important changes have occurred in the epi-demiology of ulcer disease. Already more than forty years ago, a decline in the in-cidence of peptic ulcer was observed. The discovery of H.pylorihad a further major impact on the incidence of ulcer disease. Our aim was to evaluate the trends in the incidence of ulcer disease in the Netherlands.

Methods From a computerized endoscopy database of a district hospital, the data of all patients who underwent upper gastrointestinal endoscopy in the years 1996-2005 were analyzed. Incidences of duodenal and gastric ulcers with and without complications were compared over time.

Results Overall, 20,006 upper gastrointestinal endoscopies were performed. Duo-denal ulcers were diagnosed in 696 (3.5%) cases, with signs of bleeding in 158 ulcers (22.7%): 45 (6.5%) of these ulcers were classified as Forrest I, 113 (16.2%) as Forrest II.

Gastric ulcers were diagnosed in 487 (2.4%) cases, with signs of bleeding in 60 ulcers (12.3%). Forrest I was diagnosed in 19 patients (3.9%) and Forrest II in 41 (8.4%) patients.The incidence of gastric ulcers was stable over time, while the incidence of duodenal ulcers declined.

Conclusions The incidence of duodenal ulcer disease is steadily decreasing over time in the Dutch population. With a further declining prevalence of H.pylori, the incidence of gastric ulcers is likely to exceed the incidence of duodenal ulcer in the very near future, bringing us back to a similar situation as at the beginning of the last century.

7

98

Chap ter 7

Introduction

Over the past decades, important changes have occurred in the epidemiology of ulcer disease. A peak incidence was seen among those born at the end of 19th cen-tury. The peak for duodenal ulcers followed the gastric ulcer peak with a 10-20 years lag.1 Already more than forty years ago, a decline in the incidence of peptic ulcer was suggested, based on a cohort analysis showing that the mortality due to pep-tic ulcer decreased in successive birth cohorts.2 This observation was supported by decreasing hospitalization rates for peptic ulcer over the past decades in the USA as well as in many European countries.3-4

The discovery in the early eighties of Helicobacterpylori(H.pylori)as major cause of peptic ulcer disease had a significant impact on the treatment of ulcer disease. This significant impact led to the award of the 2005 Nobel Prize in Medicine to Robin Warren and Barry Marshall. H. pylori eradication therapy was proven to cure pa-tients with previous chronic recurrent ulcer disease. Further research into the epi-demiology of H.pylori infection showed that the prevalence of this bacterium was decreasing over time in recent decades, presumably as a result of improvements in living conditions.5-6 A recent Asian study showed that the prevalence of peptic ulcer disease, mainly duodenal ulcers, was reduced in association with a decreas-ing trend in the prevalence of H.pylori, confirming that H.pylori is involved in the development of especially duodenal ulcers.7

With decreasing incidence of ulcer disease the incidence of ulcer complications might be affected as well. A recent study in the Netherlands showed that the inci-dence of the most important complication, ulcer bleeding, remained however stable in the period of 1993 – 2003, notwithstanding decreasing incidence of peptic ulcers.8 Mortality and rebleeding rates for peptic ulcer bleeding did not change either.9 In-vestigators from England reported that the incidence of gastric and duodenal ulcer haemorrhage in fact even increased among older patients. This rise was shown to be correlated with oral anticoagulant, NSAID’s and oral corticosteroid intake.10-11

Introduction of an endoscopy database allowed closer investigation of incidence and epidemiology of gastric and duodenal ulcers, complications rates and classifi-cations. We aimed to define changes in the incidence of gastric and duodenal ul-cer disease and its complications over a ten-years period in a large uninvestigated dyspeptic population in the Netherlands. Moreover, use of the database enabled us to describe the incidence of ulcer complications in more detail over time in a large group of peptic ulcer patients for the first time, by use of the Forrest classification.12

99

Incidence of duodenal and gastric ulcers

Methods

All gastrointestinal endoscopies performed from June 1996 to January 2006 in the Ikazia Hospital in Rotterdam, the Netherlands, were evaluated. The Ikazia Hospi-tal is a district hospital with an open-access endoscopy unit, where about 40% of the patients is referred by general practitioners. The large percentage of referrals by GPs from both the urban agglomeration and rural sites makes the investigated popula-tion a cross section of the Dutch population. In July 1995 a computerized endoscopy report system, Endobase III®, developed by Olympus Software, was introduced and from June 1996 onwards each gastrointestinal endoscopy performed in the cen-tre was recorded with this system. Four experienced endoscopists performed all endoscopy procedures recorded in the database. The system enables standardized methods for report writing.13 All data are coded with a specific gastrointestinal endoscopic terminology code (GET-C) which is an extension of the International Classification of Diseases 10th version (ICD-10).14

All codes in the database starting with K25 and K26, coding for gastric and duodenal ulcer, were included in this study. Furthermore, a search was performed for the word ‘ulcer’ in all reports of upper gastrointestinal endoscopies in the database. These reports were reviewed for the diagnosis of gastric and duodenal ulcers. Many patients with a gastric ulcer underwent a follow-up gastroscopy after a few weeks. These follow-up endoscopies were excluded from the analysis; so all included cases were newly diagnosed ulcers in a previously uninvestigated dyspep-tic population.

Patient characteristics such as age, gender and indication for the endoscopy were retrieved from the database. Furthermore, the Forrest classification and H.pylori status were investigated. All found ulcers were classified according to the Forrest classification.12 From 2001 onwards, the program enabled entering the For-rest classification into the database, which was done for all peptic ulcers. Ulcers entered before 2001 were re-classified using the Forrest classification based on the reports and available endoscopic images.

The H. pylori status was assessed with a rapid urease test (CLO-test) on gastric biopsy specimens.

Data were analysed in Statistical Package for the Social Sciences (SPSS), version 11.5 (SPPS, Chicago, IL). Regression of incidence of ulcers over the years was plotted together with the observed incidence of ulcers in percentage per year. Data are de-scribed by means and standard deviations (SD) or 95% confidence intervals (CI). Two-sided p values < 0.05 were considered to be statistically significant. Student’s t-test and linear regression analysis were used to compare groups and linear relations.

100

Results

In total 20,006 upper gastrointestinal endoscopies were accessible for analysis, 10,784 (53.9%) of which were performed in women. The mean age of the patients was 55.6 years with a standard deviation (SD) of 17.9 years. The total number of en-doscopies performed yearly, increased gradually over the 10 years from 1800 to 2200 per year.

Duodenal ulcers were found in 696 (3.5%) of the investigated patients, gastric ulcers in 487 (2.4%) patients. The majority of ulcer patients was male, 57.6% of duo-denal and 51.5% of gastric ulcer patients.

Duodenal ulcer patients were significantly younger than gastric ulcer patients. The mean age of duodenal ulcer patients was 57.4 years (95% confidence interval (CI) 55.9 – 58.9), gastric ulcer patients had a mean age of 66.3 years (95% CI 64.9 – 67.7) (p< 0.001). The distribution over age groups is shown in Figure 1. From 1996 to 2006 the mean age of the ulcer patients increased; for duodenal ulcer patients by 0.992 years per year (95% CI 0.44 – 1.4) (p<0.001); for gastric ulcers patients by 0.274 years per year, but not statistically significant (95% CI – 0.34 – 0.89) [Figure 2].

Classification according to the Forrest system [Table 1] demonstrated that gas-tric ulcers showed signs of bleeding (Forrest 1 and 2) less frequently than duodenal ulcers (p < 0.001). Forrest-3 ulcers (no signs of recent bleeding) were seen in 77.3% of

Chap ter 7

Figure 1 Ulcersindifferentagegroups

Age groups

Incidence(%)

0

1

2

3

4

5

6

7

8

9

10

11

12

14

>9080-8970-7960-6950-5940-4930-3920-2910-190-9

Duodenal ulcers

Age-related proportionof patients with ulcer disease

Gastric ulcers

101


Figure 2 Changeinthemeanageofduodenalandgastriculcerpatients

Year

Age

0

10

20

30

40

50

60

70

80

90

100

200620052004200320022001200019991998199719961995

Linear regression of mean ageof duodenal and gastric ulcerpatients in 10-year period.

Duodenal ulcer

Gastric ulcer

Total Population

Themeanageoftheduodenalulcerpatientsincreasedwith0.992yearsperyear(95%CI0.44-1.4)(p<0.001)comparedtoanon-significantincreaseof0.274yearsperyearforgastriculcerpatients(95%CI–0.34-0.89)

Table 1 Numberandclassificationofduodenalandgastriculcers

Duodenal Ulcers Gastric Ulcers

Number Percentage Number Percentage

Forrest1a 15 2.2 1 0.2

Forrest1b 30 4.3 18 3.7

Forrest2a 39 5.6 14 2.9

Forrest2b 37 5.3 7 1.4

Forrest2c 37 5.3 23 4.7

Forrest3 538 77.3 424 87.7

Total 696 100.0 487 100.0

NumberofduodenalulcersandgastriculcersaccordingtoForrestclassification

102

the duodenal ulcers and 87.0% of the gastric ulcers. Active bleeding (Forrest 1) was seen in 6.5% of all duodenal ulcers and 3.9 of all gastric ulcers and signs of bleeding, i.e. a visible vessel (Forrest 2a), an overlying clot (Forrest 2b) or a hematin-covered base (Forrest 2c) were diagnosed in 16.2% of the duodenal ulcers and 8.4% of the gas-tric ulcers.

Complicated ulcers, defined as Forrest 1 and 2, were more frequently seen in elderly patients, with a mean age of bleeding duodenal ulcer patients of 69.7 years (95% CI 66.8 – 72.7) versus 54.3 years (95% CI 52.8 – 55.9) (p<0.001) for non-bleeding duode-nal ulcer patients. For bleeding gastric ulcer patients the mean age was 71.6 years (95% CI 67.6 – 75.5), compared to 65.6 years (95% CI 64.1 – 67.1) for uncomplicated ulcer disease (p = 0.077). Of the patients with complicated ulcers, 60.4% and 63.4%, for duodenal and gastric ulcers respectively, were older than 70 years of age. The gender distribution was not different for various Forrest-classified ulcers, nor was the gen-der distribution in patients with complicated ulcers different from that in patients with uncomplicated ulcers. For both groups the majority of patients were males (57.7% and 54.5%, respectively; p = 0.395).

In patients with complicated ulcer disease H. pylori infection was assessed in 22.6%, with a positive result in 37.1% of the tested patients.

Chap ter 7

Figure 3 Regressionofincidenceofulcersover10yearperiod(%ofallendoscopies)

Percentage

2,0

2,5

3,0

3,5

4,0

4,5

5,0

5,5

200620052004200320022001200019991998199719961995

Percentage DU

Percentage GU

Trendsintheincidenceofduodenal(DU)andgastriculcer(GU)disease

103

The incidence of gastric ulcers remained stable over the ten years observation period. In contrast, duodenal ulcers showed a decreasing incidence, with a linear regres-sion coefficient of -.169 (p= 0.040) [Figure 3]. This decline was due to a decline in the incidence of uncomplicated ulcers only. The incidence of complicated duodenal and gastric ulcers increased over this 10-year period, as shown in Figure 4. This increase was more explicit for duodenal ulcers.

The rapid urease test (CLO-test) on biopsy specimens taken during endoscopy for the presence of H.pylori was performed in 20.5% of gastric ulcer patients and 36.4% of the duodenal ulcer patients. In 34.0% of the tested gastric ulcer patients and 56.5% of the tested duodenal ulcer patients the test results were positive for H.pylori. The overall incidence of CLO-positive tests declined significantly from 57.7% in the pe-riod 1996-2000 to 45.1% in the period 2001-2006 (p=0.02).

Discussion

The investigated population consists of patients referred for endoscopy, for a major part based on either serious recurring complaints, alarm symptoms or older age, according to the 1996 Dutch General Practitioners’ Guideline on Dyspepsia.15 Thus, the investigated population is not a cross section of the total population and in-cidences from this investigation are not fully comparable with findings concern-ing total populations. Trends over time are however clear. We found a decreasing incidence of duodenal ulcer and a stable incidence of gastric ulcer. This suggests a further amplification of the trend observed from 1980 onwards, implying a de-creasing incidence of both gastric and duodenal ulcer, with only a slight decline for gastric ulcer incidence.8 Our data suggest that the former, already slighter, decline in gastric ulcer incidence has stopped by now, whereas duodenal ulcer incidence is still further decreasing. This leads to a situation in which, similar to 100 years ago, gastric ulcer again becomes more common than duodenal ulcer, whereas for the past 100 years, duodenal ulcer disease has been more common than gastric ulcer disease in Western countries.1

Other trends that we found were an increasing incidence of complicated ulcers for both duodenal and gastric ulcers, an increasing age at diagnosis for patients with duodenal ulcers and a declining incidence of H.pyloripositive ulcers.

The overall decline of peptic ulcer disease is likely to be due to a combination of factors, including the introduction of acid suppressive medication, a decreasing prevalence of H.pylori in subsequent birth cohorts, and the development of eradica-tion treatment for H.pylori-positive ulcer patients preventing chronic relapsing ulcer disease. The introduction of newer NSAID’s and a tendency for prescription of lower doses of aspirin for patients with cardiovascular disease may also have contributed to the changing epidemiology of ulcer disease.


104

The Forrest classification can predict the risk of re-bleeding from peptic ulcer disease and is therefore an important tool to select patients for endoscopic treat-ment.16-17 In this study a minority, 12.3% of the gastric and 22.7% of the duodenal ul-cers, were complicated by bleeding. As in other studies10 bleeding ulcers were more frequently seen in older patients and in men.

While we observed a decreasing incidence of ulcers, there was a rise in the pro-portion of complicated ulcers, an observation in line with data presented in other studies.8-9 The incidence of complicated ulcers rose in the investigated period in both gastric and for a greater part in duodenal ulcers [Figure 3 and 4]. The decline of duodenal ulcers in particular can even be attributed completely to a decrease of uncomplicated Forrest 3 ulcers.

The changes in the proportional distribution of ulcers of complicated versus non-complicated ulcers in the investigated period may be influenced by more accurate description of ulcers in endoscopic reports in the recent years. The endoscopy re-port program Endobase III®, which was used for this study, enabled more detailed standardized report writing as a result of improvements in the program from 2001 onwards. So, in the updated version of the program the need for free text was minimized. This may have influenced the description of the ulcer classification. We support the suggestion from other publications showing a stable incidence or slight increase of ulcer bleeding compared to overall peptic ulcer disease. It was suggested that the use of NSAID’s are the main cause of this change.8, 10-11 The high propor-tion of negative CLO-tests in bleeding ulcers supports the hypothesis that bleeding ulcers are more often related to NSAID use, although it may also partly be due to low sensitivity of this test in upper gastrointestinal bleeding patients.18 Adherence to guidelines to prevent NSAID-related upper gastrointestinal events is low in the Netherlands as well as in the USA.9, 19 This might be an important cause of the opposite trends in the incidence of complicated versus uncomplicated peptic ulcer patients.

Both ulcer types show an increase in incidence rate with growing age. Duodenal ulcers occur in our population at a younger age than gastric ulcers. The incidence of duodenal ulcers starts to rise at about 20 years of age, in contrast with 50 years of age for gastric ulcers. This again indicates an etiological factor for duodenal ulcers early in life, such as infection with H.pylori. These findings are not completely com-parable to other studies which show an older age of duodenal ulcer patients than in this study.10

In 1996 an update of the Dutch General Practitioners Guidelines of 1993 on Dys-pepsia was published.15 In this 1996 guideline it was advised to diagnose H.pylori status by endoscopy biopsy, as only H.pylori infection in patients with ulcers should be treated. The role of H.pylori in other dyspeptic symptoms was stated as unclear.

In 2004 a new multidisciplinary guideline for dyspepsia was published in the

Chap ter 7

105


Figure 4a Trendsintheincidenceofcomplicatedanduncomplicatedduodenalulcers(%ofallendoscopies)

Percentage

0,0

0,5

1,5

2,5

3,0

3,5

4,0

4,5

200620052004200320022001200019991998199719961995

1,0

2,0

DU no complications*

DU complications†

Trendsintheincidenceofcomplicatedanduncomplicatedulcers* Duodenalulcerswithoutcomplicationsofbleeding(Forrest3)† Duodenalulcerswithcomplicationsofbleeding(Forrest1and2)

Figure 4b Trendsintheincidenceofcomplicatedanduncomplicatedgastriculcers(%ofallendoscopies)

Percentage

0,0

0,5

1,0

2,0

2,5

3,0

200620052004200320022001200019991998199719961995

1,5

GU no complications*

GU complications†

Trendsintheincidenceofcomplicatedanduncomplicatedulcers* Gastriculcerswithoutcomplicationsofbleeding(Forrest3)† Gastriculcerswithcomplicationsofbleeding(Forrest1and2)

106

Netherlands by the quality institute for health services CBO and the Dutch general practitioners society.20 In this guideline, an H.pylori test is recommended if there are persistent non-reflux-like complaints without alarm symptoms. If H. pylori is negative, treatment with PPI for 2-4 weeks is recommended. If symptoms persist or diagnostic certainty is desired, endoscopy is recommended. Because these guide-lines were published in 2004 and the total number of endoscopies performed was still rising since this year, this guideline is not suspected to have influenced the incidence and complications of peptic ulcer disease in this study. However, it is pos-sible that noninvasive testing for H pylori and the use of trials of therapy for dyspep-tic patients have had a significant effect on the rate of performance of endoscopy in duodenal ulcer patients, resulting in an apparent decline in incidence.

A CLO-test was performed only in a small proportion of the patients. Other tests like the urea C¹³ breath test or serology may have been used in patients included in this study, in particular in case of bleeding. However those test results were not included in the endoscopy database and thus could not be analysed for this study.

From the results obtained we conclude that the incidence of uncomplicated peptic ulcer disease, especially for duodenal ulcers, declines in our Western popu-lation. This will soon bring us back to a situation where gastric ulcers are more common than duodenal ulcers. The incidence of complicated ulcer disease is how-ever rising, which underlines the need to be more alert in providing patients using NSAID’s with adequate gastroprotection to prevent ulcer disease and its complica-tions.

Chap ter 7

107

References

1 Cucino C, Sonnenberg A. The long-term time trends of peptic ulcer and ulcerative colitis are interrelated. AmJGastroenterol. 2002;97(10):2657-62.

2 Susser M, Stein Z. Civilisation and peptic ulcer. Lancet. 1962;1:115-9.3 Sonnenberg A. Temporal trends and geographical variations of peptic ulcer disease. Ali-

mentPharmacolTher. 1995;9 Suppl 2:3-12.4 Sonnenberg A. Time trends of ulcer mortality in Europe. Gastroenterology. 2007;132(7):2320-

7.5 Kuipers EJ, Pena AS, van Kamp G, Uyterlinde AM, Pals G, Pels NF, et al. Seroconversion for

Helicobacter pylori. Lancet. 1993;342(8867):328-31.6 Loffeld RJ, van der Putten AB. Changes in prevalence of Helicobacter pylori infection in

two groups of patients undergoing endoscopy and living in the same region in the Nether-lands. ScandJGastroenterol. 2003;38(9):938-41.

7 Xia B, Xia HH, Ma CW, Wong KW, Fung FM, Hui CK, et al. Trends in the prevalence of peptic ulcer disease and Helicobacter pylori infection in family physician-referred unin-vestigated dyspeptic patients in Hong Kong. AlimentPharmacolTher. 2005;22(3):243-9.

8 Post PN, Kuipers EJ, Meijer GA. Declining incidence of peptic ulcer but not of its compli-cations: a nation-wide study in The Netherlands. AlimentPharmacolTher. 2006;23(11): 1587-93.

9 van Leerdam ME, Vreeburg EM, Rauws EA, Geraedts AA, Tijssen JG, Reitsma JB, et al. Acute upper GI bleeding: did anything change? Time trend analysis of incidence and outcome of acute upper GI bleeding between 1993/1994 and 2000. AmJGastroenterol. 2003;98(7):1494-9.

10 Higham J, Kang JY, Majeed A. Recent trends in admissions and mortality due to pep-tic ulcer in England: increasing frequency of haemorrhage among older subjects. Gut. 2002;50(4):460-4.

11 Weil J, Langman MJ, Wainwright P, Lawson DH, Rawlins M, Logan RF, et al. Peptic ulcer bleeding: accessory risk factors and interactions with non-steroidal anti-inflammatory drugs. Gut. 2000;46(1):27-31.



14 Groenen MJM, Hirs W, Becker H, Kuipers EJ, van Berge Henegouwen GP, Fockens P, et al. Gastrointestinal Endoscopic Terminology Coding (GET-C): A WHO-approved extension of the ICD-10. DigDisSci. 2006.

15 Numans ME, de Wit NJ, Geerdes RHM, Muris JWM, Starmans R, Postema PJ. NHG-Stan-daard Maagklachten (M36). In: Thomas S, Geijer RMM, van der Laan JR, Wiersma T, edi-tors. NHG-Standaarden voor de huisarts II. Utrecht: Wetenschappelijke uitgeverij Bunge; 1996. p. 211-23.

16 Cook DJ, Guyatt GH, Salena BJ, Laine LA. Endoscopic therapy for acute nonvariceal upper gastrointestinal hemorrhage: a meta-analysis. Gastroenterology. 1992;102(1):139-48.

17 Laine L, Peterson WL. Bleeding peptic ulcer. NEnglJMed. 1994;331(11):717-27.18 Gisbert JP, Abraira V. Accuracy of Helicobacter pylori diagnostic tests in patients with

bleeding peptic ulcer: a systematic review and meta-analysis. AmJGastroenterol. 2006; 101(4):848-63.


108

19 Abraham NS, el Serag HB, Johnson ML, Hartman C, Richardson P, Ray WA, et al. National adherence to evidence-based guidelines for the prescription of nonsteroidal anti-inflam-matory drugs. Gastroenterology. 2005;129(4):1171-8.

20 Kwaliteitsinstituut voor de Gezondheidszorg CBOeNHG. Multidisciplinaire Richtlijn Maagklachten: Van Zuiden Communications B.V.; 2004.

Chap ter 7

109



Trends in the diagnostic yield

of open access

upper gastrointestinal endoscopy

Submi t t ed


ErnstJ.Kuipers2

RobJ.Th.Ouwendijk1

Chap ter 8

111

Abstract

Purpose We intended to compare the diagnostic yield of upper endoscopy proce-dures in patients referred by general practitioners, specialists and clinical refer-rals, in order to establish the effectiveness of open access endoscopy over time.

Methods Cumulative 10-year data of an open access endoscopy unit were analyzed. Data of all patients referred for their first diagnostic upper endoscopy were ana-lyzed, comparing diagnostic yield between different referring groups and trends over time.

Results In 21,441 examinations eligible for analysis, 42% patients were referred by general practitioners, 38% by specialists and 20% were clinical referrals. The propor-tion of first line referrals increased from 35 to 45% (p<0.001). Pathology was found in 74% of the endoscopies, without a difference in diagnostic yield between refer-ral groups. The proportion of positive endoscopies increased over the investigated period.

Conclusions The high yield of diagnostic upper endoscopies in all referral groups warrants open access upper gastrointestinal endoscopy.

8

112

Chap ter 8

Introduction

Open access endoscopy is defined as the performance of endoscopic procedures re-quested by referring physicians without prior clinical consultation of the endosco-pist.1 This form of referral for endoscopy is extensively available in both the United States and Europe.2 Open access endoscopy is commonly offered for esophago-gas-tro-duodenoscopy (EGD), flexible sigmoidoscopy, and in some centers for colono-scopy.

Open access upper gastro-intestinal endoscopy has had a major impact on the workload for endoscopy units in the last two decennia.3 A rise in number of en-doscopies of 25% was seen from 1999 to 2004 in the Netherlands, leading to a total of 1137 upper endoscopies per 100.000 persons per year with a mean waiting time of 3.1 weeks between request and procedure.4 The rise in upper endoscopies led to discussion about possible overuse of diagnostic endoscopy capacity. Several studies published between 1997 and 2001 have shown that endoscopies indicated by gas-troenterologists had higher diagnostic yields than those requested by non-gastro-enterologists.5-8 There are however only limited data on the findings of endoscopic procedures in patients referred by general practitioners in an open access endoscopy system.1 Furthermore, little is known about trends in diagnostic yield of open access upper gastrointestinal endoscopy over time in different referral groups that might have been caused by the decreasing prevalence of Helicobacterpylori over the last decades, the widespread use of proton pump inhibitors since their introduc-tion in the late 1980’s and introduction of guidelines for gastroprotection in NSAID users since the late 1990’s. If these measures should have led to a decrease in the yield of open access endoscopy, it can be questioned whether open access procedures are justified in view of scarcity of endoscopic capacity. We therefore studied the diagnostic yield of upper gastrointestinal endoscopy in an open access endoscopy unit in three different referring groups. Findings in patients referred by general practitioners (first line) were compared with patients referred by specialists (second line) and patients referred during hospital admission (clinical patients). The trends in incidence of diagnoses in a ten-year period were assessed, with special emphasis on reflux esophagitis as a common disorder in all age and referral groups.

Methods

The Ikazia hospital in Rotterdam, the Netherlands, offers an open access endoscopy service, including esophagogastroduodenoscopy (EGD), to both general prac-titioners and hospital medical staff serving a population of approximately 150.000 persons. Data from all consecutive upper gastrointestinal endoscopies performed

113

Trends in the diagnostic yield of upper endoscopy

in our centre from July 1996 until July 2006 was included in the study. Endoscopic examinations performed in this clinic are recorded in a computerized endoscopy report system, Endobase III®, developed by Olympus Software, since June 1996, af-ter a 12-month development period. By linking the Endobase system and the hospi-tal information system (HIS), patient characteristics are imported from the HIS and the report and digital endoscopic images are sent to the HIS.

Four experienced endoscopists performed the endoscopies and used the program with the standardized reports and text blocks for description of the endoscopic findings.9 All methods of report writing are automatically coded with a specific gastrointestinal endoscopic terminology code (GET-C), developed as an extension of the International Classification of Diseases 10th version (ICD-10).10 In this way all endoscopic data were collected in a similar way and stored within the database.

We compared the diagnostic yield of the endoscopies referred by general prac-titioners versus those referred by medical staff from the outpatient clinic and pa-tients from within the hospital. Only patients who underwent the examination for the first time were included. All follow-up and surveillance endoscopies for specific conditions such as Barrett’s esophagus, celiac disease and ulcer disease were left out of the analysis.

Demographics and details of the diagnoses were recorded. If the diagnosis re-flux esophagitis was made, a classification was made; before 2000 esophagitis was classified according to the Savary-Miller classification (grades I - IV)11 and from 2000 onwards according to the LA-classification (grades A-D).12 Before 2000 reflux esophagitis was re-classified. For comparison, the Savary-Miller scores were recod-ed in LA classification grades. Although these classifications cannot be compared completely, we chose to recode the Savary-Miller scores I to IV consecutively into LA grades A to D.

The data were exported from Endobase to be evaluated statistically in Statistical Package for the Social Sciences (SPSS), version 14.0 (SPPS, Chicago, IL). Regression of incidence of diagnosis over the years was plotted in percentage per year. Data are described by means and 95% confidence intervals (95% CI). Two-sided p values < 0.05 were considered to be statistically significant. Linear regression analysis were used to compare groups and linear relations.

All findings were classified into the following specific groups: normal, esopha-gitis, hiatal hernia, gastritis, duodenitis, gastric ulcers, duodenal ulcers, esopha-geal neoplasm, gastric neoplasm, duodenal neoplasm, varices, therapeutic inter-ventions, and ‘miscellaneous’.

Diagnoses with clinical consequences (including cancer, esophagitis, hiatus hernia of great dimension, varices, ulcers, erosive gastritis and/or duodenitis,

114

esophagitis and Barrett’s esophagus) were considered to be ‘major pathology’, other positive endoscopic diagnoses (mild gastritis or duodenitis, small hiatal hernia) were considered as ‘minor pathology’.

In the ten year study period time-trends for different diagnoses were investi-gated.

Results

A total of 22,268 esophago-gastro-duodenoscopies (EGD) were performed by 4 endo-scopists from July 1996 until July 2006. In total 827 endoscopies were excluded from the analysis because they were follow-up endoscopies for Barrett’s esophagus (575), celiac disease (13), ulcer disease (129) or malignancy (110). Thus, the data of 21,441 examinations were eligible for analysis.

Of all patients, 42% were referred by a general practitioner, 38% by a specialist, and 20% were patients referred while being admitted to the hospital [Table 1]. Over the years, the proportion of patients referred by general practitioners increased from 35 to 45% with a linear regression coefficient of .615 (p=0.036); the proportion of referrals by specialists and the proportion of referrals at hospital admission declined significantly [Figure 1].

Chap ter 8

Figure 1 Regressionofproportionofreferralstoopen-accessendoscopyunitintime.

Percentage

15

20

25

30

35

40

45

50

20062005200420032002200120001999199819971996

First Line

Second Line

Clinical

TrendsindistributionofpatientsreferredbyFirstLine(generalpractitioner),SecondLine(specialist)andClinicalpatients.

115

Of all patients referred for EGD a small majority (54%) was female, which was more pronounced in the specialist referrals [Table 1]

The mean age of all patients was 55.7 years. Patients referred by general practitio-ners were significantly younger with a mean age of 51.2 (95% CI 50.9 – 51.6) years compared with 54.7 (95% CI 54.3 – 55.1) years for specialist referrals and 67.1 (95% CI 66.6 – 67.6) years for admitted patients (p<0.05 for first and second line versus clini-cal patients) [Figure 2]. This age difference was related with different diagnosis distributions between referral groups.


Table 1 Patientcharacteristicsindifferentreferralgroups

First line referrals Second line referrals

Clinical referrals Total

Number (%) Number (%) Number (%) Number (%)

Numberofendoscopies 8971(41.8) 8197(38.2) 4273(19.9) 21441

Gender MaleFemale

4297(47.9)4674(52.1)

3492(42.6)4705(57.4)

2018(47.2)2255(52.8)

9807(45.8)11634(54.2)

Meanage 51.2Range(14-96)

54.7Range(0-96)

67.1Range(8-104)

55.7Range(0-104)

Figure 2 Agedistributioninpatientsreferredbygeneralpractitioners,byspecialistsandathospitaladmission.

Age groups

%

0

5

10

15

20

25

30

>9080-8970-7960-6950-5940-4930-3920-2910-190-9

First Line

Referred by

Second Line

Clinical Patients

116

Pathology was found in 15,820 patients (74%). The referral groups differed slight-ly but significantly with respect to the prevalence of pathology, with the highest diagnostic yield (75%) in the first line referrals, 73% in the specialist referrals, and 74% in the admitted patients (p = 0.001). Females were more likely to have a normal endoscopy (30%) than male patients (22%) (p < 0.005). The diagnostic yield of endo-scopy increased in the last ten years from 59% to 79% (p < 0.0001) for the comparison of first versus last year).

The first line and second line referrals both showed an increase over time in posi-tive endoscopies both with major and minor pathology. Endoscopies performed in hospital-admitted patients showed an increase in pathology with clinical conse-quences, but a stable prevalence of minor pathology [Figure 3].

Table 2 shows the most frequent endoscopic findings for the different referral groups and the total population. In 4943 (23%) of the endoscopies, patients had more than one diagnosis during their first endoscopy procedure, with two diagnoses in 4037 patients (19%), three diagnoses in 811 patients (4%), and four diagnoses in 95 patients (0.4%).

Chap ter 8

Figure 3 Trendsofuppergastrointestinalendoscopydiagnoses(negative,positivewithmajorpathologyorpositivewithminorpathology)overatenyearsperiodinpatientsreferredbygeneralpractitioners,byspecialistsandathospitaladmission.

%

10

20

30

40

50

60

70

80

20061996

Year

Clinical PatientsFirst line Second line

20061996 20061996

Major Pathology

Negative Endoscopy

Minor Pathology

117

Reflux esophagitis was seen in 16% of all patients, 55% of these were grade A, 24% grade B, 12% grade C, 5% grade D and 4% grade D with complications. Reflux esopha-gitis was more frequently seen in first line-referred patients than second line and clinical patients and no clear trend was observed over time [Figure 4]. Reflux esophagitis was diagnosed more frequently in men, this predominance was more pronounced for grades A and B than for grades C and D esophagitis [Figure 5].

Milder types of reflux esophagitis were most prevalent in age groups of 30-59 years.Newly diagnosed Barrett’s esophagus was seen in 4% of the patients, rising with

age [Figure 6]. There were no significant differences between the referring groups with respect to the prevalence of Barrett’s esophagus.

Upper gastrointestinal malignancies were found in 387 patients (2%), 205 pa-tients were diagnosed with adenocarcinoma of the stomach and 156 patients were diagnosed with esophageal cancer [Table 2]. The proportion of patients who were


Table 2 Incidenceofdiagnosesindifferentreferringgroups

Diagnosis First Line(8971)

Second Line(8197)

Clinical Patients(4273)

Total(21441)

n % n % n % n %

Normal 2250 25.1 2257 27.5 1114 26.1 5621 26.2

HiatalHernia 2507 27.9 2009 24.5 644 15.1 5160 24.1

Esophagitis 1867 20.8 1171 14.3 444 10.4 3482 16.2

Gastritis 2236 24.9 1751 21.4 646 15.1 4633 21.6

Duodenitis 900 10.0 573 7.0 340 8.0 1813 8.5

GastricUlcer 169 1.9 148 1.8 240 5.6 557 2.6

DuodenalUlcer 319 3.6 148 1.8 286 6.7 753 3.5

Barrett’sesophagus 380 4.2 328 4.0 116 2.7 824 3.8

NeoplasmEsophagus 69 0.8 39 0.5 48 1.1 156 0.7

NeoplasmStomach 60 0.7 49 0.6 96 2.2 205 1.0

NeoplasmDuodenum 2 0.0 11 0.1 13 0.3 26 0.1

Varices 16 0.2 76 0.9 99 2.3 191 0.9

Therapeuticinterventions 153 1.7 760 9.3 525 12.3 1438 6.7

Miscellaneous 961 10.7 1070 13.1 685 16.0 2716 12.7

118

Chap ter 8

Figure 4 Incidenceofrefluxdiseaseindifferentreferringgroupsduringatenyearsperiod.

Percentage

0

5

10

15

20

25

20062005200420032002200120001999199819971996

Year

Esophagitis First Line

Second Line

Clinical Patients

Figure 5 TrendsinincidenceofrefluxesophagitiswithLAclassificationformalesandfemales.

Gender

%

0

2

4

6

8

10

12

16

20061996

Year

M

20061996

F

14

Reflux grade A

Reflux grade B

Reflux grade C

Reflux grade D

119


Figure 6 IncidenceofrefluxesophagitiswithLAclassificationandnewdiagnosedBarrett’sesophagusfordifferentagegroups.

Percentage

0

2

4

8

6

12

>9080-8970-7960-6950-5940-4930-3920-2910-190-9

Age groups

10

Barrets mucosa

Reflux grade A

Reflux grade B

Reflux grade C

Reflux grade D

Reflux grade Dwith complications

Figure 7 Trendsinincidenceofesophagealandgastricneoplasmsformalesandfemales.

%

0,00

0,25

0,50

0,75

1,00

1,25

1,75

20061996

Year

M

20061996

F

1,50

Gender EsophagealNeoplasm

GastricNeoplasm

120

diagnosed with cancer was the highest in the hospital group, accounting for 4% of all the endoscopies performed in this group, but proportions in patients referred by the first line and the second line were not different (1.2 and 1.5%). Cancer of the esophagus and stomach were more frequently diagnosed in men than in women, without a significant change over time [Figure 7].

In 2716 patients (13%) miscellaneous conditions were diagnosed, either as main or as additional diagnosis, including angiodysplasia, achalasia, conditions related to previous surgery, diverticula, Mallory-Weiss lesions, other causes of esophagitis, stenosis, Schatzki rings and webs in the esophagus.

Therapeutic interventions were performed in 1438 patients (7%), including dila-tations of peptic stenosis and anastomotic strictures after surgery, balloon dilata-tion for achalasia, as well as haemostasis in bleeding ulcers, treatment of bleeding and non-bleeding varices, placement of endoprostheses in the esophagus, place-ment of gastrostomy catheters and placement of feeding tubes in the duodenum and jejunum.

Discussion

Open access endoscopy has led to increasing numbers of upper gastrointestinal en-doscopy procedures and longer waiting times, with the proportion of upper endo-scopies referred by general practitioners increasing by about 10% to 45% of all refer-rals in the investigated hospital.

It has been suggested that open access endoscopy might shorten the time to en-doscopy, and thus enable diagnosis of upper gastrointestinal malignancies in an earlier state with a better survival.13 This was however not confirmed in other stud-ies, and the usefulness of first line referral was doubted. Open access endoscopy did even lead to longer waiting times for all referred groups.3, 14

In the investigated endoscopy practice patients could be referred by general practitioners from 1988 onwards. From 1996 onwards data were collected in an En-dobase system, which allowed comparison between groups and over time. By doing so, we found that the proportion of referrals by general practitioners increased and the proportion of clinical patients and referrals by specialists declined in the last ten years.

With an increase of first line referred patients a reduction in diagnostic yield of overall performed endoscopies was hypothesized. Overall the diagnostic yield of upper endoscopy was high with only 26% of the investigations without any abnor-malities. This yield of upper endoscopy is higher than in other studies.6-8, 15-16 In the past the diagnostic yield was described to be higher in specialist referred patients.8

Chap ter 8

121

Our results show a remarkable rise of diagnostic yield in all referral groups, sug-gesting that the selection for this procedure has improved. In both first and sec-ond line referrals the positive endoscopies with clinical consequences accounted for about 45-50% of all endoscopies. Besides, endoscopies with minor pathology and even normal endoscopies can influence the subsequent management of patient, avoiding unnecessary therapies.17-18 Malignancies were seen more frequently in clinical patients than in other referred patients. As malignancies were seen more frequently in the older population, this might explain the stronger association be-tween clinical referrals and cancer. Another explanation is that complications of malignancies like bleeding and obstruction are reasons for endoscopy at hospital admission.

The male-female ratio of patients is comparable to that in other studies. No clear explanation was found for the surprising higher proportion of female patients in the group referred by specialists. The age of patients was different between the three referring groups with hospital patients being almost 15 years older than pa-tients referred by general practitioners. This has a major impact on the distribution of diagnoses between the three referral groups.

In this article we concentrate on reflux esophagitis. Reflux esophagitis is common and was diagnosed in a great proportion of the patients (16%). Although women and men are reported to have generally similar patterns of endoscopic severity in gastroesophageal reflux diseases,19 we found reflux disease to be more frequent and more serious in men than in women. The highest incidence of reflux esophagitis was seen in the patients referred by general practitioners. This could be explained by the fact that patients referred by the general practitioner probably used less acid suppressive medication or for a shorter period than patients referred by a specialist or from out of the hospital. The highest incidence of reflux esophagitis was seen in the patients referred by general practitioners. Reflux esophagitis is a condition that can be treated by the general practitioner himself, after endoscopic confirmation of the diagnosis. If a general practitioner should suspect another diagnosis, the patient would rather be referred to the specialist and not for first line endoscopy. In the investigated ten years period, the incidence of reflux disease increased in all referring groups and for all grades of severity.

Grade A reflux esophagitis was diagnosed most frequently, accounting for about 55% of all the newly diagnosed reflux esophagitis patients. The incidence of esopha-gitis declined from 2004 onwards, especially in the first line referrals [Figure 4]. This was seen especially in the mildest grades of reflux (A and B) [Figure 5]. After an initial rise in reflux esophagitis, the number stabilized and even declined. this might have to do with a discussion, ending in changed Dutch Guidelines for gen-eral practitioners on the diagnosis and treatment of upper GI symptoms in 2004,20


122

implying a proton pump inhibitor (PPI) test treatment with for 2-4 weeks is recom-mended in suspected gastroesophageal reflux disease without alarm symptoms. If reflux symptoms persist or diagnostic certainty is desired, endoscopy is recom-mended. Due to this guideline a higher proportion of referred patients will use PPIs before their first endoscopy. PPI showed in a meta-analysis a healing rate of 11.7% per week.21 This may explain the declining incidence of endoscopically proven esophagitis in the last 3 years. In this short period of 2 to 4 weeks particularly mild esophagitis will be reduced. Adherence to gastroprotective agents in NSAID’s users increased from 11.3% in 1996 to 44.6% in 2005 in a study performed in the same area as this population, probably leading to a decrease in esophagitis.22

The peak incidence of the mildest reflux esophagitis grades (A and B) around 30-59 years and the raising incidence of grade C and D with rising years of age are in correspondence with other studies.23

Newly diagnosed cases of Barrett’s esophagus were seen in 4% of all patients. The incidence of Barrett’s esophagus increased gradually with growing age [Figure 6]. This can be explained by the fact that development of Barrett’s esophagus is the result of longstanding reflux of gastric content into the esophagus.24

Our data show that the use of a standardized report system with an extended cod-ing system is suitable for analyzing trends in endoscopic data. The diagnostic yield is high in all investigated referring groups. Despite discussion in the past about incorrect referral and the rise of proportion of referrals by general practitioners for upper gastrointestinal endoscopy it seems that the yield of first line diagnostic up-per endoscopy is comparable with specialist referrals. We conclude that the claim of the first line on the scarcity of endoscopic capacity is still justified and that this service of open access endoscopy should be continued.

Chap ter 8

123

References

1 Eisen GM, Baron TH, Dominitz JA, Faigel DO, Goldstein JL, Johanson JF, et al. Open access endoscopy. GastrointestEndosc. 2002;56(6):793-5.

2 Silcock J, Bramble M. Open access gastroscopy: second survey of current practice in the United Kingdom. Gut. 1997;40(2):192.

3 Paterson H, McCole D, Auld C. Impact of open-access endoscopy on detection of early oesoph-ageal and gastric cancer 1994-2003: population-based study. Endoscopy. 2006;38(5):503-7.

4 Terhaar Sive Droste J, Craanen M, Kolkman J, Mulder C. dutch endoscopic capacity in the era of colorectal cancer screening. Neth J Med. 2006;64:371–3.

5 Charles RJ, Chak A, Cooper GS, Wong RC, Sivak MV, Jr. Use of open access in GI endoscopy at an academic medical center. GastrointestEndosc. 1999;50(4):480-5.

6 Rossi A, Bersani G, Ricci G, Defabritiis G, Pollino V, Suzzi A, et al. ASGE guidelines for the appropriate use of upper endoscopy: association with endoscopic findings. Gastrointest–Endosc. 2002;56(5):714-9.

7 Trevisani L, Sartori S, Gilli G, Chiamenti CM, Gaudenzi P, Alvisi V, et al. Appropriate-ness of upper gastrointestinal endoscopy: a hospital-based study. Dig Dis Sci. 2001;46(12): 2695-9.

8 Zuccaro G, Jr., Provencher K. Does an open access system properly utilize endoscopic re-sources? GastrointestEndosc. 1997;46(1):15-20.


10 Groenen MJM, Hirs W, Becker H, Kuipers EJ, van Berge Henegouwen GP, Fockens P, et al. Gastrointestinal Endoscopic Terminology Coding (GET-C): A WHO-approved extension of the ICD-10. DigDisSci. 2006:1004-8.

11 Savary M, Miller G. The esophagus. Handbook and atlas of endoscopy: Verlag Gassman, Solothurn, Switzerland; 1978.

12 Armstrong D, Bennett J, Blum A, Dent J, De Dombal F, Galmiche J, et al. The endoscopic assessment of esophagitis: a progress report on observer agreement. Gastroenterology. 1996;111(1):85-92.

13 Blackshaw G, Barry J, Edwards P, Allison M, Lewis W. Open-access gastroscopy is associ-ated with improved outcomes in gastric cancer. European Journal of Gastroenterology & Hepatology. 2003;15(12):1333.

14 Spahos T, Hindmarsh A, Cameron E, Tighe M, Igali L, Pearson D, et al. Endoscopy waiting times and impact of the two week wait scheme on diagnosis and outcome of upper gastro-intestinal cancer. Postgraduate medical journal. 2005;81(961):728.

15 Adang RP, Vismans JF, Talmon JL, Hasman A, Ambergen AW, Stockbrugger RW. Appropri-ateness of indications for diagnostic upper gastrointestinal endoscopy: association with relevant endoscopic disease. GastrointestEndosc. 1995;42(5):390-7.

16 Loffeld RJ, van der Putten AB. The yield of UGIE: a study of a ten-year period in the Zaanstreek. NethJMed. 2003;61(1):14-8.

17 Charles RJ, Cooper GS, Wong RC, Sivak MV, Jr., Chak A. Effectiveness of open-access endos-copy in routine primary-care practice. GastrointestEndosc. 2003;57(2):183-6.

18 Hungin A, Thomas P, Bramble M, Corbett W, Idle N, Contractor B, et al. What happens to patients following open access gastroscopy? An outcome study from general practice. The British Journal of General Practice. 1994;44(388):519.


124

19 Lin M, Gerson L, Lascar R, Davila M, Triadafilopoulos G. Features of gastroesophageal reflux disease in women. The American journal of gastroenterology. 2004;99(8):1442-7.

20 De Wit N, van Barneveld T, Festen H, Loffeld R. Richtlijn Maagklachten. Ned Tijdschr Geneeskd. 2005;149:1386-92.

21 Chiba N, Gara C, Wilkinson J, Hunt R. Speed of healing and symptom relief in grade II to IV gastroesophageal reflux disease: a meta-analysis. Gastroenterology-Orlando. 1997;112(6):1798-810.

22 Van Soest E, Sturkenboom M, Dieleman J, Verhamme K, Siersema P, Kuipers E. Adherence to gastroprotection and the risk of NSAID related upper gastrointestinal ulcers and haem-orrhage. Alimentary pharmacology & therapeutics. 2007;26(2):265-75.

23 Lööf L, Götell P, Elfberg B. The Incidence of Reflux Oesophagitis a Study of Endoscopy Re-ports from a Defined Catchment Area in Sweden. Scandinavian journal of gastroenterol-ogy. 1993;28(2):113-8.

24 van Soest E, Dieleman J, Siersema P, Sturkenboom M, Kuipers E. Increasing incidence of Barrett’s oesophagus in the general population. Gut. 2005;54(8):1062.

Chap ter 8

125

Summary and discussion

Samenvatting en discussie

Chap ter 9

127

Summary and Discussion

Gastrointestinal endoscopic investigations with flexible endoscopes started in the 1960. Since that time endoscopic investigations changed from an innovative tech-nique to visualize the gastrointestinal tract to a routine investigation for patients with complaints of the gastrointestinal tract. Besides diagnostic investigations endoscopy is progressively used for therapeutic interventions and screening, espe-cially colorectal carcinoma screening.

A report has to be made from the investigation to inform the referring doctor about the findings, therapy and advice. Besides this it is important to know in forthcoming endoscopy the findings en therapy of former investigations. Report-ing is therefore an essential part of the endoscopic investigation. As an imaging method with numerous repetitive manoeuvres as well as findings, gastrointestinal endoscopy reports are particularly suitable for electronic storage and processing.

Standardization and coding of endoscopic reports makes it suitable for research, and quality measurements. Quality indicators that are in development will con-tain aspects of endoscopic reporting or items that can be retracted from the data-base that is build by the report system.

To be able to retract data from the database of the report system it is important to have standardized reporting and database building. Fur this purpose a working group was formed, called Trans.it, with 3 university and seven district hospitals. The aim of this working group was to create uniformity and consensus on gastro-intestinal endoscopic reporting. The second purpose of this working group was to establish a coding system for gastrointestinal endoscopic findings and terminology to be able to analyze data anonymously. To create a report system that is suitable for every endoscopists we generated three different ways of report writing. These dif-ferent ways can be chosen independently after each examination.

Firstly standard reports were created, consisting of a complete report, describing the results of the whole investigation based on one or a combination of diagnoses. Secondly, we constructed text blocks, by which a report is built up selecting differ-ent text-blocks for separated parts or different diagnoses. With the combination of these text-blocks a complete report is generated. Thirdly, the newest way to com-pose a report is using structured data entry. In this way a sentence and subsequent-ly a report is built up selecting different aspects of a finding in a specific part.

In all these different ways allowed for reporting of endoscopic findings accord-ing to international classifications. The medical context of the reporting was dis-cussed in the Trans.it working group and consensus was obtained. The three dif-ferent ways of report writing are coded in the same way en stored in the database.

9

128

Chap ter 9

In this way forthcoming new quality measurements can commonly be retracted from the system.

In Chapter 2 the first two ways of report writing are described and discussed. Standard reports are mainly used to report examinations without abnormalities and examinations with frequently seen abnormalities. In cases of more rare find-ings or a combination of diagnoses the use of text blocks is more suitable. With these two ways of report writing, 90 % of the reports can be composed within a short period of time. For the remaining part structured data entry is advised. This way of report writing describes the findings very punctual, but is more complex and takes more time.

The implementation of new techniques is often criticized. Change from hand-written to standardized computerized reporting is generally thought to require more time for report composing. Physicians often insist to keep their usual way of working, especially when a new way of working not directly results in obvious gain of time or quality aspects. We therefore studied in Chapter 3 time aspects of three different ways of report writing. Computerized reports were compared to dictated reports and handwritten reports. The time needed to compose the report and send it to the referring physician was in computerized reports 86 seconds, compared to handwritten reports 113 seconds and dictated reports 65 seconds with additional time for the typist of 172 seconds. This showed that standardized computerized re-porting consumed similar time than the other methods. The main advantage of computerized reporting is the possibility to capture and store endoscopic images and the structured way of storing the data of the endoscopy performed for forth-coming research, quality indicators or management information.

Starting with computerized systems demands an investments in hardware, and software including construction of necessary links between the system and the hospital information system. One of these links is a connection between the report system and the financial department. Computerized systems are known to record investigations more accurate, leading to billing benefits. Besides direct fi-nancial benefits computerized system also showed to improve quality aspects. In Chapter 4 a study on costs for the above mentioned three different ways of report writing (computerized, dictated and handwritten) was performed. Computerized reporting was shown to be the most expensive way of reporting because of high initial costs. After five years, at the moment that the initial investments are de-preciated, the costs per report were comparable with those of the two other ways of report writing. In a cost-benefit analysis already after three years the computerized system show a positive financial benefit.

In the computerized system, besides standard reports and text blocks also struc-tured data entry is allowed. For every examination the endoscopist is free to choose one of the three ways of report writing. To be able to compare the results of the

129


investigations, the reports need to be coded in a similar way. For this reason a cod-ing system had to be developed. It was chosen to extend a widely used coding sys-tem, the International Classification of Diseases, 10th version, of the World Health Organization (WHO). In Chapter 5 this extension is described. The WHO defined rules to extend the ICD-10. In this way adaptations are possible to be coded and in the same time the generated code is retraceable to its original ICD-10 code. With the extension made in this Gastrointestinal Endoscopic Terminology Code (GET-C) for gastrointestinal endoscopic findings and terms it is possible to generate codes for all items that are recorded during endoscopy. With new developed techniques and diagnosis, this GET-C is consecutive adapted to correctly code new endoscopic terms. These data can anonymously be exported to another database to compare data between different endoscopy units and countries.

The automatically recorded codes are stored in the database of each hospital. Within the Trans.it project research is being performed on the anonymous data. Therefore a central database was created with the anonymous data of all investiga-tions in the participating hospitals. Within the project it was agreed to use stan-dard reports and text blocks although the report maker could use free text or adjust the report. Obviously, this could potentially influence the correctness of the auto-mated terminology coding. In Chapter 6 a multicentre study is described to assess to what extent automatically generated GET-C codes are in agreement with the ac-tual text of the endoscopy report. A total of 500 colonoscopy reports were selected randomly on a total of 2500 examinations from 5 different hospitals and evaluated on indication, endoscopy findings, therapeutic interventions. Indications were correctly coded in 92% of the reports. Correct coding of endoscopic findings varied between 42 and 93% per hospital. This great variation was explained by incorrect use of the reporting system, in particular the selection of a standard report and consecutive adding or deleting diagnosis by free text, resulting in a incorrect code. The incorrectness of the coding was also scored for clinical relevance, resulting in 89% of all the investigations to be clinically correctly coded. These result warrant further improvement in requirements for composing reports and training of endo-scopists for using the system to enhance the correctness of the database. We intend to induce better recording of endoscopy data by showing each hospital and individ-ual endoscopists their data, compared to the anonymous results of other hospitals. The result of this benchmarking has be evaluated in the future by a new study on the correctness of coding in the Trans.it group.

In the Ikazia Hospital, Rotterdam, the computerized system was used from 1996 onwards. It was possible to study the endoscopic data over a ten year period. Trends in different diagnoses were analyzed. In Chapter 7, we studied the incidence of gastric and duodenal ulcer disease over time. In a 10 year period a total of 20,006 endoscopies were performed in previously uninvestigated dyspeptic patients. Gas-

130

tric ulcers were diagnosed in 2.4% of the cases. Duodenal ulcers were diagnosed in 3.5 % of the patients and this incidence declined in this 10 year period. A further decline of duodenal ulcer incidence with a stable incidence of gastric ulcer disease will soon lead to a situation similar to 100 years ago in which gastric ulcer is more common than duodenal ulcer disease in the Netherlands. The overall decline of ul-cer disease is attributed to be due to several factors, including widespread use of acid-suppressive medication, a decreasing prevalence of H.pylori and eradication of this bacterium in new ulcer patients, thus preventing recurrence.

In Chapter 8 different referring groups and their yield of upper gastrointesti-nal endoscopy were compared over time. In an open access endoscopy unit patients were referred by general practitioners, specialist and clinical referrals. In total 21,441 first diagnostic upper endoscopies were eligible for analysis. The proportion of patients referred by their general practitioner raised from 35 to 45% in 10 years. Abnormalities were found in the majority (74%) of the investigation without differ-ences in diagnostic yield between the referral groups. The percentage of patients with abnormalities increased over the period.

Gastro-esophageal reflux diseases was the most common diagnosis, observed in 16% of the patients. Men were affected more often and had more severe grades of esophagitis than women. The highest prevalence of reflux disease was seen in patients referred by general practitioners. This may be explained by the fact that these patients used less acid suppressive medication or for a shorter period than patients referred by a hospital physician. After an initial rise in reflux esophagitis, the number stabilized and even declined. In 2004 a new guideline for general prac-titioners was published on the diagnosis and treatment of upper gastrointestinal symptoms. This guideline recommended use of proton pump inhibitor (PPI) for 2-4 weeks as initial approach to patients with new onset dyspepsia. If reflux symptoms persist, endoscopy is recommended. This can be the explanation of a decline in in-cidence of reflux diseases in the last years of this study.

In the past there has been discussion about incorrect referral and the rise in re-ferrals by general practitioners. Our data show a high diagnostic yield in all refer-ral groups, justifying open access endoscopy in our setting.

Conclusions

In this thesis different studies are described on standardization and coding of gas-trointestinal endoscopy reports. Good report writing of endoscopic investigations is of major importance for the treatment, screening and follow-up of patients. With standardization, reports will fulfill all recommended items for complete endoscopy reports.

Chap ter 9

131

New computerized medical systems request a great initial financial investment for the program as well as for the connection with other hospital systems. Besides these financial investments, efforts has to be made by users of the system. There-fore the report system has to be quick, easy to learn, and suitable for use in busy daily practice. Studies in this thesis showed comparable costs and even financial benefits after initial investments and also a quick system compared to other ways of report writing.

Because of automatically coding of recorded findings the developed system cre-ates a extensive database suitable for research. The created GET-C coding system facilitates in anonymously recording of these data. At the same time items appro-priate for quality indicators can be or are already stored in the system. The system has showed to be capable to investigate diagnostic yield and trends of endoscopic examinations described in this thesis.

In the future these systems become essential for monitoring of quality, quality improvement, and management of endoscopy units.


133

Samenvatting en discussie

In 1960 werd gestart met gastrointestinale endoscopische onderzoeken met flexi-bele endoscopen. Sinds die tijd zijn endoscopische onderzoeken veranderd van een techniek om het maag-darmkanaal in beeld te brengen, in een routine onderzoek bij patiënten met maag- en darmklachten. Naast diagnostische onderzoeken wordt endoscopie in toenemende mate gebruikt voor therapeutische ingrepen en screening, van voornamelijk dikke darmkanker.

Van het endoscopisch onderzoek moet een verslag worden gemaakt om de ver-wijzende arts te informeren over de bevindingen, de verrichte therapie en het ad-vies. Daarnaast is het belangrijk om de bevindingen en therapie van eerdere on-derzoeken in de toekomst opnieuw te kunnen inzien. Rapportage is daarmee een essentieel onderdeel van het endoscopisch onderzoek. Als beeldvormend onderzoek met vaak voorkomende bevindingen en verrichtingen zijn gastrointestinale endo-scopie verslagen bijzonder geschikt voor elektronische vastlegging en verwerking.

Standaardisatie en codering van endoscopische verslagen maakt het geschikt voor onderzoek en tevens voor kwaliteitsmetingen. Kwaliteitsindicatoren die in ontwikkeling zijn zullen aspecten bevatten van endoscopische verslaglegging. Daarnaast kunnen sommige indicatoren worden verkregen uit de database die wordt opgebouwd door het verslagleggingsysteem.

Om gegevens op de juiste manier te kunnen achterhalen uit de database van het verslagleggingsysteem is het van belang om een gestructureerde verslaglegging en opbouw van deze database te hebben. Hiervoor werd een werkgroep gevormd, Trans.it genaamd, met 3 universitaire ziekenhuizen en zeven perifere ziekenhui-zen. Het doel van deze werkgroep is het creëren van uniformiteit en consensus voor gastrointestinale endoscopische verslaglegging. Het tweede doel van deze werk-groep is om een codering systeem te ontwikkelen voor gastrointestinale endoscopi-sche bevindingen en terminologie zodat gegevens anoniem opgeslagen en geana-lyseerd kunnen worden. Om een systeem te ontwikkelen dat voor elke endoscopist makkelijk te gebruiken is, zijn er drie verschillende vormen van verslaglegging ge-creëerd binnen het gebruikte programma. Deze verschillende vormen van verslag-legging kunnen onafhankelijk van elkaar, door de endoscopist bepaald, gekozen worden na afloop van het onderzoek.

Ten eerste zijn standaard verslagen gecreëerd, bestaande uit een volledig ver-slag, waarin de resultaten van het hele onderzoek op basis van een of een combi-natie van diagnosen is samengesteld. Daarnaast de mogelijkheid om een verslag te maken met tekstblokken. Hiermee wordt een verslag opgebouwd door het selecte-ren van verschillende tekstblokken corresponderend met verschillende diagnosen en verschillende anatomische delen van het onderzoek. Ten derde, de laatst ont-

134

wikkelde manier om een verslag samen te stellen, met behulp van gestructureerde data invoer. Op deze manier wordt een zin uit het verslag opgebouwd door het se-lecteren van een specifieke endoscopische bevinding, waarvan vervolgens verschil-lende items dienen te worden geselecteerd.

Bij al deze verschillende vormen van verslaglegging, worden de bevindingen be-schreven volgens internationaal geaccepteerde classificaties van gastrointestinale endoscopische afwijkingen. De medische inhoud van deze verslagen wordt bespro-ken binnen de Trans.it werkgroep en hierover wordt consensus verkregen. De drie verschillende vormen van verslaglegging worden op dezelfde manier gecodeerd en in de database vastgelegd. Op deze manier is het mogelijk om alle onderzoeken te vergelijken, kwaliteitsindicatoren te kunnen meten en eventueel toekomstige kwaliteitsmetingen uit het systeem te halen.

In hoofdstuk 2 worden de eerste twee manieren van verslaglegging beschreven en besproken. Standaardverslagen worden voornamelijk gebruikt om onderzoeken zonder afwijkingen of onderzoeken met veel voorkomende afwijkingen te beschrij-ven. In het geval dat er zeldzamere afwijkingen of een combinatie van meerdere diagnosen worden gevonden, is het gebruik van tekstblokken beter geschikt. Met deze twee vormen van verslaglegging kan 90 % van de verslagen in een korte pe-riode worden samengesteld. Voor de overblijvende onderzoeken wordt gestructu-reerde invoer geadviseerd, die veel nauwkeuriger kan worden samengesteld, maar daardoor ook meer keuzes vraagt en meer tijd kost.

De implementatie van nieuwe technieken krijgt vaak kritiek op de werkvloer. Bij het veranderen van verslaglegging van handgeschreven of gedicteerde verslagen naar gestandaardiseerde verslagen met behulp van een computer, is een vaak ge-hoorde klacht, angst voor de toenemende tijdsinspanning van de endoscopist. Artsen willen hierbij graag hun gebruikelijke manier van werken behouden. Vooral wanneer de nieuwe manier van werken niet direct resulteert in een con-crete winst van tijd of kwaliteitsaspecten. We hebben daarom in hoofdstuk 3 de tijdsaspecten van drie verschillende manieren van verslaglegging onderzocht en beschreven. Gecomputeriseerde verslagen werden vergeleken met gedicteerde en handgeschreven verslagen. De tijd die nodig is om het rapport samen te stellen en te versturen naar de verwijzende arts was in gecomputeriseerde rapporten 86 se-conden, in vergelijking met handgeschreven rapporten 113 seconden en gedicteerde verslagen 172 seconden, waarvan 65 seconden tijd voor de typiste nodig was om een verslag te maken en te versturen. Dit maakt het samenstellen van het verslag met een gestandaardiseerd computer verslagleggingsystemen vergelijkbaar met andere vormen van verslaglegging. Het belangrijkste voordeel van gecomputeriseerde ver-slaglegging is de mogelijkheid om het verslag op ieder gewenst moment te kunnen inzien en de mogelijkheid om endoscopische beelden op te slaan en terug in te zien.

Chap ter 9

135

Samenvat ting en discussie

Daarnaast levert het gestructureerd opslaan van de gegevens van de endoscopie on-derzoeken mogelijkheden voor wetenschappelijk onderzoek, kwaliteitscontrole en management informatie.

Het starten met gecomputeriseerde verslagsystemen vraagt een grote financiële investering voor hardware, software en verschillende koppelingen tussen het ver-slagsysteem en het ziekenhuis informatie systeem. Gecomputeriseerde systemen bewijzen wel onderzoeken nauwkeuriger te registreren, waardoor er facturerings-voordelen optreden en andere kwaliteitsaspecten verbeteren. In hoofdstuk 4 wordt een studie beschreven over de kosten van de hierboven genoemde drie verschillende manieren van de verslaglegging (gecomputeriseerde, gedicteerde en handgeschre-ven verslaglegging). Een kosten-batenanalyse werd uitgevoerd tussen gedicteerde en gecomputeriseerde verslaglegging. Gecomputeriseerde verslaglegging blijkt de duurste manier van rapporteren te zijn vanwege de hoge investeringskosten. Na vijf jaar, als een groot deel van de investeringskosten zijn afgeschreven zijn de kos-ten per verslag vergelijkbaar met de twee andere manieren van verslaglegging. In een kosten baten analyse wordt al na drie jaar van het gecomputeriseerde systeem een positief financieel voordeel gezien.

In het gecomputeriseerde systeem, is naast standaard verslagen en tekstblok-ken, ook gestructureerde data invoer mogelijk. Na ieder endoscopisch onderzoek is de endoscopist vrij om een van de drie manieren van de verslaglegging te kie-zen. Om de op verschillende manier vastgelegde resultaten te kunnen vergelijken, moeten de verslagen gecodeerd worden op een identieke manier. Hiervoor is een coderingssysteem ontwikkeld. Er werd gekozen om een veel gebruikt coderingsys-teem, de International Classification of Diseases, 10e versie, van de World Health Organization (WHO) uit te breiden. In hoofdstuk 5 wordt deze extensie beschreven. De WHO heeft regels gedefinieerd om de ICD-10 uit te breiden. Op deze manier zijn aanpassingen en uitbreidingen van de codes mogelijk waardoor specifieker geco-deerd kan worden maar tegelijkertijd de uitgebreide code eenvoudig te traceren is tot de oorspronkelijke ICD-10 code. Met de uitbreiding specifiek voor gastrointes-tinale endoscopische bevindingen en terminologie is het mogelijk om alle items die beschreven worden tijdens een endoscopie te coderen. Met de ontwikkeling van nieuwe technieken en de beschrijving van nieuwe diagnosen, wordt deze GET-C aangepast om nieuwe endoscopische termen te kunnen coderen. Deze gegevens kunnen anoniem worden geëxporteerd naar een andere database om gegevens tus-sen verschillende endoscopie afdelingen te vergelijken.

De automatisch gekoppelde codes worden opgeslagen in de database van elk ziekenhuis. Binnen het Trans.it project zal onderzoek worden uitgevoerd op de anonieme endoscopische gegevens. Daarvoor is een centrale database gemaakt met de anonieme gegevens van alle endoscopische onderzoeken van de deelnemende ziekenhuizen.

136

Binnen het project is overeengekomen om de gestandaardiseerde vormen van verslaglegging (standaard verslagen, tekstblokken en gestructureerde invoer) te gebruiken. Het is in het programma tevens mogelijk om vrije tekst te gebruiken en het verslag na de selectie van de gestandaardiseerde verslaglegging voor uit-eindelijke verzenden aan te passen. Gebruik van vrije tekst en aanpassingen na de selectie van gestandaardiseerde verslaglegging zou invloed kunnen hebben op de juistheid van de automatisch gegenereerde codering. In hoofdstuk 6 wordt een multicenter studie beschreven die is uitgevoerd om te beoordelen in hoeverre auto-matisch gegenereerde GET-C codes in overeenstemming zijn met de feitelijke tekst van het endoscopie verslag. In totaal werden 500 colonoscopie verslagen at random geselecteerd uit een totaal van 2500 onderzoeken van 5 ziekenhuizen en werden deze codes en verslagen geëvalueerd op de indicatie, de endoscopie bevindingen en therapeutische interventies. Indicaties waren correct gecodeerd in 92% van de rapporten. Correcte codering van de bevindingen varieert tussen 42 en 93% per zie-kenhuis. Deze grote variatie is te verklaren door onjuist gebruik van het systeem om een verslag samen te stellen, met name de selectie van een standaard verslag en vervolgens het verwijderen of toevoegen van bevindingen in het uiteindelijke verslag, resulterend in een onjuiste code. De onjuistheid van de codering werd eveneens gescoord op klinische relevantie met als resultaat dat in 89% van alle on-derzoeken correct klinisch werd gecodeerd. Deze resultaten maken het nodig om richtlijnen voor het samenstellen van verslagen verder te verbeteren en daarnaast endoscopisten aanvullend bij te scholen voor het op de juiste manier gebruiken van het verslagleggingsysteem. Hierdoor kan de betrouwbaarheid van de database nog verder verbeteren. Hiervoor zullen verschillende ziekenhuizen en individuele en-doscopisten hun gegevens gespiegeld krijgen aan de gemiddelde resultaten van de andere ziekenhuizen binnen de werkgroep. De resultaten van deze benchmarking zullen in de toekomst door een nieuwe studie over juistheid van codering binnen de Trans.it groep worden geëvalueerd

In het Ikazia Ziekenhuis, Rotterdam, wordt het gecomputeriseerde systeem ge-bruikt vanaf 1996. Het was mogelijk om de endoscopische gegevens over een peri-ode van tien jaar te onderzoeken. Trends in de verschillende diagnoses werden ge-analyseerd. In Hoofdstuk 7 worden de incidentie van maag - en duodenum ulcera beschreven in de tijd. In een periode van 10 jaar werden in totaal 20.006 endosco-pieën uitgevoerd bij eerder niet onderzochte patiënten met dyspeptische klachten. Maag ulcera werden gediagnosticeerd bij 2,4% van de patiënten. Duodenum ulcera werden gediagnosticeerd bij 3,5% van de patiënten en deze incidentie daalde in deze periode van 10 jaar. Met een verdere daling van duodenum ulcera en tegelijkertijd een stabiele incidentie van maag ulcera verwachten we in een situatie terug te ko-men waarin, vergelijkbaar met 100 jaar geleden in Nederland, maag ulcra weer va-ker voorkomen dan duodenum ulcera. De totale daling van ulcera is waarschijnlijk

Chap ter 9

137

Samenvat ting en discussie

te wijten aan verschillende factoren, waaronder zuuronderdrukkende medicijnen, een daling van de prevalentie van H. pylori en de behandeling van deze bacterie.

In hoofdstuk 8 worden de opbrengsten van oesofago-gastro-duodensocopieën van verschillende verwijzende groepen vergeleken in de tijd. In de onderzochte open acces endoscopie afdeling werden patiënten polklinisch verwezen door huis-artsen en specialisten en daarnaast een groep klinische verwijzingen. In totaal kwamen 21.441 primaire diagnostische oesofago-gastro-duodenoscopieën in aan-merking voor het onderzoek. Het aandeel van de patiënten verwezen door hun huisarts steeg van 35 tot 45% in 10 jaar tijd. Afwijkingen werden gevonden in de meerderheid (74%) van onderzoeken zonder significante verschillen in de diagnos-tische opbrengst tussen de verschillende verwijzende groepen. Het percentage pa-tiënten met afwijkingen steeg in de onderzochte periode.

Vooral gastro-oesofageale reflux ziekte werd gediagnosticeerd in een groot deel van de patiënten (16%). Mannen hadden vaker en ernstigere vormen van reflux-ziekte dan vrouwen. De hoogste incidentie van reflux ziekte werd waargenomen bij patiënten die werden verwezen door de huisarts. Dit kan worden verklaard door het feit dat deze patiënten minder zuurremmende medicatie gebruikten of deze medicatie voor een kortere periode gebruikten dan patiënten verwezen door een specialist of vanuit het ziekenhuis. Na een aanvankelijke toename in de incidentie van reflux ziekte, stabiliseerde en daalde deze incidentie in de laatste jaren van de studie. In 2004 werd een nieuwe huisartsenrichtlijn uitgebracht voor de diag-nostiek en behandeling van gastro-intestinale symptomen. In deze richtlijn wordt een behandeling van 2-4 weken met een protonpompremmer (PPI) aanbevolen. Als reflux- symptomen aanhouden, wordt endoscopie aanbevolen. Dit kan een verkla-ring zijn voor de daling van de incidentie van reflux ziekte in de laatste jaren van deze studie zijn.

Enkele studies hebben de mogelijkheid van direct verwijzen voor een endoscopie in een openacces endoscopieafdeling bekritiseerd. Het zou leiden tot onjuiste ver-wijzing en stijging van het aantal verwijzingen voor een onderzoek waarvoor, door het tekort aan MDL artsen, reeds een wachttijd bestaat.

Onze onderzoeksgegevens tonen een hoge diagnostische opbrengst in alle ver-wijzende groepen, waardoor het in stand houden van een openacces endoscopie afdeling gerechtvaardigd is.

Conclusie

In dit proefschrift worden de verschillende studies beschreven over standaardisatie en codering van gastro-intestinale endoscopieverslagen. Goede verslaglegging van endoscopische onderzoeken is van groot belang voor de behandeling, het onder-

138

zoek en de follow-up van patiënten. Met standaardisatie, zullen de verslagen alle geadviseerde punten voor complete endoscopieverslagen kunnen vervullen. Nieu-we geautomatiseerde medische systemen vragen een grote financiële investering voor het programma evenals voor de verbinding met andere het ziekenhuissyste-men. Naast deze financiële investeringen, moeten gebruikers van het systeem in-spanningen leveren. Daarom moet het rapportsysteem snel, gemakkelijk te leren, en geschikt zijn voor gebruik in een drukke dagelijkse praktijk. De studies in dit proefschrift toonden vergelijkbare kosten en zelfs financiële voordelen na aanvan-kelijke investeringen en ook een snel systeem in vergelijking met andere manieren van verslaglegging.

Door het automatisch coderen van vastgelegde bevindingen bouwt het ontwik-kelde systeem een uitgebreid database met endoscopie gegevens op, geschikt voor research. Het gecreëerde GET-C coderingsysteem maakt anonieme vastlegging van gegevens mogelijk. Tegelijkertijd kunnen gegevens benodigd voor kwaliteitsindi-catoren worden opgeslagen in het systeem. Het systeem heeft aangetoond om op-brengst en trends van endoscopische onderzoeken te kunnen onderzoeken. In de toekomst worden deze systemen essentieel voor kwaliteitscontrole, kwaliteitsver-betering, en management van endoscopieafdelingen.

Chap ter 9

139

Dankwoord

Curriculum Vitae

appendix

141

Dankwoord

Uithoudingsvermogen, moet je hebben om een promotie tot een goed einde te bren-gen. Maar niet alleen uithoudingsvermogen voor de promovendus. Zeker ook voor iedereen die direct of indirect aan de totstandkoming van dit proefschrift heeft bij-gedragen. Ik ben daarvoor velen dankbaar, waarvan ik een aantal met name wil noemen.

Graag wil ik mijn promotor Ernst Kuipers bedanken. Beste Ernst, jij hebt zeker uithoudingsvermogen moeten hebben. Bijna 10 jaar geleden heb ik de mogelijk-heid gekregen om bij jou in opleiding te komen tot MDL arts. Ik ben je hiervoor nog steeds dankbaar. Vanaf dat moment ben jij ook als promotor betrokken geraakt bij dit proefschrift. Je analytische en kritische benadering heeft dit proefschrift naar een hoger niveau gebracht. Je was altijd beschikbaar voor vragen en commentaar en ik ben erg onder de indruk van de wijze waarop en snelheid waarmee je weer een nieuw manuscript beoordeelde en van advies voorzag. Ik hoop dat na de afronding van dit proefschrift we op meerdere fronten kunnen blijven samen werken.

Heel bijzonder wil ik mijn co-promotor Rob Ouwendijk bedanken. Onze eerste con-tacten gaan ver terug. Als co-assistent interne geneeskunde ben ik in 1995 voor het eerst bij je gekomen in het Ikazia ziekenhuis. Ik heb daar ook voor het eerst ken-nis gemaakt met de afdeling endoscopie waardoor ik erg geïnspireerd werd. Aan het eind van mijn co-schappen zijn wij gestart met een onderzoeksproject. Hier is de eerste basis gelegd voor dit proefschrift. In eerste instantie is een elektronische vorm voor verslaglegging gecreëerd met het programma Endobase in het Ikazia ziekenhuis. Om deze te standaardiseren en te uniformeren hebben we het initia-tief genomen tot het starten van het Trans.it project.

Ik dank je voor je uithoudingsvermogen en je stimulerende en enthousiastme-rende begeleiding. Er waren momenten dat jij degene was die mij aan het werk bleef houden, altijd steunend en altijd geloof houdend in mij. Ik hoop dat de ge-sprekken die we hadden en de vriendschap die daaruit ontstaan is in de toekomst zullen blijven bestaan en dat we nog veel zullen samenwerken ondanks de afstand tussen ons beider werken woonlocatie.

Graag wil ik de leden van de leescommissie, Prof. dr. H.W. Tilanus, Prof. dr. C.H.J. van Eijck en Prof. J.F.W.M. Bartelsman, en de leden van de grote commissie bedan-ken voor hun bereidheid zitting te nemen in mijn promotiecommissie.

Een zeer belangrijk onderdeel bij de totstandkoming van dit proefschrift is het werk van de Trans.it groep. In eerste instantie wil ik hiervoor Gerard van Berge

142

Henegouwen bedanken. Vanaf het eerste uur behoor jij tot de initiatiefnemers van dit vernieuwende project op gebied van endoscopie verslaglegging in Nederland. Direct was je zeer enthousiast en een stimulator van zowel het project als de basis van dit proefschrift. Na jou emeritaat ben je steeds betrokken en actief gebleven binnen deze groep waarvoor veel dank.

Voor de ondersteuning van het Trans.it project ben ik dank verschuldigd aan Jans-sen-Cilag en Olympus Nederland. Beide bedrijven hebben met meerdere personen ondersteuning gegeven aan het slagen van dit project. Met name wil ik noemen, Otto Leensma, Mark Wolff en Renier van Dinter van Janssen-Cilag die tevens mij de gelegenheid hebben gegeven om een jaar volledig aan dit project te kunnen wer-ken. Daarnaast wil ik Henk Braat en Gerard van der Hoorn van Olympus Nederland bedanken. Vanaf het begin zijn jullie betrokken geweest bij Trans.it en hebben de ondersteuning gegeven voor nieuwe ideeën vanuit de groep ter verbetering van het programma Endobase. Dank voor de prettige samenwerking en ik hoop dat die nog verder kan worden doorgevoerd.

Aan het Trans.it project hebben vanuit het Ikazia meerder mensen meegewerkt. Ram Soekhoe is met een nieuwe ontwikkeling binnen Endobase gestart. Beste Ram, dank voor je samenwerking en het onderzoek dat we samen hebben mogen uitvoeren waardoor naar voren kwam dat werken met computers wel aanpassing van de gebruiker vergt, maar ook veel voordelen kan opleveren. Sanna Mulder heeft na Ram de draad opgepakt en het Trans.it project verder ondersteund. Beste Sanna, jij hebt gewonnen door je proefschrift reeds vorig jaar af te ronden waarvan een deel uit de database van Endobase kwam. Het was prettig en erg ge-zellig om met je samen te werken.

In een klein kantoortje in het Ikazia ziekenhuis waar het episch centrum van Trans.it was gelegen, heb ik veel samen gewerkt met Adrienne Zandbergen. Beste Adrienne, samen als onderzoekers in een perifeer ziekenhuis is heel bijzonder ge-weest. Jij op een heel ander gebied, maar toch ook raakvlakken. Ik dank je voor de gezelligheid en het klankbord dat je voor mij was.

Bij mijn werkzaamheden in het Ikazia heb ik veel praktisch steun gehad van het se-cretariaat. Wil en Mildred, ik wil jullie bedanken voor al die ondersteuning evenals de gezelligheid en afleiding tussen het werk door. Daarnaast wil ik het personeel van de endoscopieafdeling bedanken voor alle ondersteuning en gezelligheid. Vele uren heb ik bij jullie doorgebracht, vaak achter de computer, maar zeer regelmatig ook tijdens endoscopieën. Dank voor jullie gastvrijheid en bereidheid om geduld te hebben met weer een nieuwe versie van Endobase die moest worden uitgetest.

Dankwoord

143

Dankwoord

Ook wil het secretariaat in het Erasmus MC en met name Wendy bedanken. Wendy, tijdens mijn verblijf in het Erasmus MC en nu op afstand vanuit Arnhem heb jij de zaken rondom dit proefschrift zo geregeld dat het allemaal mooi op tijd is afgeko-men.

Het secretariaat van de MDL in het Rijnstate ziekenhuis wil ik bedanken voor de ondersteuning in de laatste vier jaar. Een proefschrift afronden naast een volledige praktijk vraagt ondersteuning op vele vlakken, die ik van jullie altijd heb gekregen. Met name Karin wil ik bedanken voor de hulp bij de laatste afrondingen van dit proefschrift, die uiteraard vaak op het korte termijn gerealiseerd moesten worden.

Mijn maten wil ik bedanken voor de ruimte en tijd die ze hebben gegeven voor de afronding van dit proefschrift.

Mijn paranimfen, Jolande en Herbert, wil ik bedanken dat zij mij deze dag willen bijstaan. Lieve Jolande, daar waar je kon heb je als zus mij bijgestaan in de rea-lisatie van dit proefschrift. Bedankt dat je zelfs vanuit het buitenland en met je voet omhoog doorging tot in de late uurtjes om mij te ondersteunen. Beste Herbert, de laatste jaren hebben we regelmatig over dit proefschrift gesproken, maar met name gingen de gesprekken met jou en je gezin ook over andere zaken. Dank dat je vandaag naast me wil staan.

Mijn ouders wil ik danken voor hun steun. Jullie hebben je tijdens mijn middelbare schooltijd en ook tijdens mijn opleiding tot arts weleens afgevraagd of het hele-maal tot een goed eind zou komen. De steun en het vertrouwen die jullie mij daar toch steeds weer in gaven, creëerde de mogelijkheid om me te ontwikkelen tot arts en onderzoeker; de basisvoorwaarden van dit proefschrift. Mijn broer Paul, wil ik bedanken voor het voorbeeld dat jij bent geweest door mij voor te gaan met jouw promotie. Dat heeft mij positief gestimuleerd.

Mijn allergrootste dank gaat uit naar Lonneke. Lieve Lonneke, de laatste tien jaar is er regelmatig een verandering van plannen geweest. Een avond die anders verliep, een vakantie die later startte, een weekend zonder mij. Op sommige momenten be-leefde jij de druk van deze promotie nog meer dan ik. Desondanks heb jij me altijd de ruimte en de steun gegeven waardoor ik het heb kunnen afronden. Ik ben je daar ontzettend dankbaar voor. De vlag mag eindelijk uit! Ik hoop dat wij samen met Sabijn, Kees-Jan en Phillip nog heel lang gelukkig zullen zijn, jongens…

‘hetboekjeisaf’.

145

Curriculum Vitae

De auteur van dit proefschrift werd op 4 december 1970 geboren in Tilburg. In 1989 behaalde hij het eindexamen Atheneum aan het Theresialyceum in Tilburg. In hetzelfde jaar ging hij geneeskunde studeren aan de Erasmus Universiteit in Rot-terdam. In 1995 behaalde hij het doctoraalexamen. Zijn co-schappen heeft hij voor een groot deel gevolgd in het Ikazia ziekenhuis te Rotterdam.

In 1997 behaalde hij cum laude het arts examen, waarna hij onderzoek ging doen op het gebied van endoscopische verslaglegging samen met dr. R.J.Th. Ouwendijk.

Zijn opleiding tot Maag-, Darm-, en Leverarts is hij gestart in december 1999, met de vooropleiding Interne Geneeskunde (opleider: Dr. R.J.Th. Ouwendijk) in het Ikazia ziekenhuis te Rotterdam, waarna hij van mei 2003 tot mei 2006 zijn oplei-ding voltooide in het Erasmus Medisch Centrum te Rotterdam (opleider: Prof. dr. E.J. Kuipers).

Van mei 2006 tot januari 2007 werkte hij als MDL-arts in het Erasmus MC en het Ikazia ziekenhuis te Rotterdam.

Vanaf januari 2007 is hij werkzaam als MDL-arts binnen de maatschap MDL van het Rijnstate ziekenhuis te Arnhem. Hij is gehuwd met Lonneke van Poppel en samen hebben zij een dochter Sabijn en twee zonen Kees-Jan en Phillip.

isbn 978-90-817036-1-1nur 876, 981

Coverdesign and Layout: Bottenheft, Marijenkampen/Arnhem

Printed by JP Offset, Duiven

© 2011 Marcel Groenen, Arnhem

All rights reserved. No part of this thesis may be reproduced, stored in a database or retrieval system, or published, in any form or in any way, electronically, mechanically, by print, photo print, microfilm or any other means without prior written permission from the author.

Documents

Standardization and Coding of Gastrointestinal Endoscopy ... Marcel Johannes Maria.pdf · Standardization and Coding of Gastrointestinal Endoscopy Reports Standaardisatie en codering